Duration 01:30:22
16+
Play
Video

Citrix Synergy TV - SYN217 - Multi-site recommended practices, reloaded

Jarian Gibson
Sr Solutions Architect at Nutanix
  • Video
  • Table of contents
  • Video
Citrix Synergy Atlanta 2019
May 22 2019, Atlanta, GA, United States
Citrix Synergy Atlanta 2019
Video
Citrix Synergy TV - SYN217 - Multi-site recommended practices, reloaded
Available
In cart
Free
Free
Free
Free
Free
Free
Add to favorites
559
I like 0
I dislike 0
Available
In cart
Free
Free
Free
Free
Free
Free
  • Description
  • Transcript
  • Discussion

About speakers

  • Jarian Gibson
    Sr Solutions Architect at Nutanix
  • Shane Kleinert
    Solutions Architect, South East at Choice Solutions
  • Kees Baggerman
    Technical Director, Engineering - End User Computing, Databases and G2K at Nutanix

About the talk

Topic: IT

Enjoyed last year’s jam-packed multi-site scenarios recommended practices session? It’s back, with an extra 45 minutes of multi-site awesomeness! Join this session for all-new content including multi-cloud scenarios, operational aspects, hypervisor concepts and deeper dives on various failover scenarios. Hosted by CTPs who have collectively worked on many multi-site builds, this session will give you the confidence and clear direction to start designing your own multi-site initiatives. We’ll cover Citrix Virtual Apps and Desktops, Citrix Files, Citrix ADC, Citrix Endpoint Management, and Citrix Cloud services. Topics include the importance of multi-cloud resiliency, flow matrix of various design decisions in a multi-site build, and failure scenarios and how to protect against them. Note: This session will be available for on-demand viewing post-event on Citrix Synergy TV.

Share

Hello, write my check. So welcome everybody today to send to 17 multisite recommended practices. So 00:04 I offered the guys to do you do an introduction but I also told them that I would totally make things up about them. So they want to introduce 00:13 themselves if they will know. My name is in lip Hook and the south of England. 00:22 My name is John Gibson. I'm a Senior Solutions architect for and you can extend the performance solutions Engineering Group focus on you. See I'm also 00:37

a citrus TTP and you can find me at Jardin Gibson on Twitter. My name is Gage Pokemon. I work in the promise ocean engineering team where I'm a 00:45 technical director at nutanix. Animoso is 242 PC everybody think liner to Choice Solutions a 00:53 Senior Solutions architect based out of the Southeast region, and I do have a face if you look at the app, I didn't submit my face or my profile. So I 01:03 guess I am real so looking forward to kick things off today. Pray, too. Well, let's move on to the agenda. It's a 90-minute session. We've got a lot 01:11

of slides a lot of ground to cover blood of demos as well. So we're not going through every single item that we're going to head because we've got 01:19 over a hundred forty slides 10-22 key subjects will want to touch upon to be prepared. We're going to do this in a 01:27 rapid Pace because we have a lot of content to share with you but these are the the key items that were going to discuss like why we're doing this 01:37 multi-site presentation. What are your typical decisions that you would see in a multi-site consideration? I do level of detail. So we'll dive into to 01:44

keep the Olympic moron on some of these components and what's going on in the real world. Obviously, we've got some housekeeping rules. It's a 01:53 90-minute session YouTube If you have to go to the restrooms, please do so. We don't want to keep you here and 02:02 is there a restroom in this one just 02:10 Also, and this is a this is important to us. This is the hashtag for our session since u17. Please leave feedback, 02:27 right? These guys have done this discussion last year and we're back based on your feedback. Apparently Citrix saw the feedback and they appreciate it 02:37

the time that these guys has ever invested in this session and we want to prove our session as well. Tell him I appreciate your your being here. If 02:46 you have feedback leave it with this hashtag and will improve our session even more since we have a lot of content will try to have questions at the 02:55 end but we don't get to the questions at the end. Feel free to grab ass after at the conference or even ping us out on Twitter as well. We have the 03:02 answer to question that we can't get to them today. Just take a second one of things that we did spend a lot of time and team here to build out a pack 03:09

of a real world scenario allowed at we did a lot of the testing and mo videos in so it's basically a multi-site available allowed. So we basically 03:17 built it we had days basement. Under his bed and left hook UK. We had a VPN site to site VPN out the Azure UK South. We had a secondary region in 03:25 the US and then we had also said we had such a regions and we also had a non-friend Citrix appointment and everything was fully highly available 03:35 through the whole stack, which will talk to you today. So looking forward to it. today if you want to talk it through with a 03:42

multi-site is so If you think about Molly saw you need to the dictionary definition of it is having or involving more than 03:52 one site. If you think about in Angie's of confusing is spitting or spreading your services across multiple access points not doesn't necessarily mean 04:02 it needs to be in a different accent or even a different country. You can have to know why you have Services split across multiple entry points within 04:12 the same height sensor some of the common at 04:21

to exit pinned and that's where you use both States census, but you designate resources to run out one of the other depending on that weasels other 04:31 considerations uses near the geographical location. That's really important because you don't want to be Can I login on in Sydney in your tires in the 04:41 Netherlands and Holland cuz it the performance bo4 when Once the monster performance goes down there about 04:51 user experience the minute you saw impaction on user experience. This is it will file from there on in. 05:01

Weather is required and then allowing that to happen. These are business considerations to start off with 05:13 two from our perspective. We start off with business considerations and work our way to stack facts about to you have uses 05:23 you've got to devices now because of the sheer 05:33 number of devices that you can access your your deployments from they need to be at to identify them and apply the right policy you access points over 05:42 the world and date. And you need to link over that together, but then we'll use honey to do if you lose diet or an access point you need 05:52

another reason or another of your deployment. So that's complex 06:01 really complex if we had a nice 06:11 house. She had a Moment of clarity. There are two things that a car engine is complexed you operate like if you if you 06:21 open up the hood and you look at the car and you think it's complex. I do right? I've got no technical background other than this stuff. So if I look 06:31 at a car engine, I think it's complex. But if you look at the car shop still have one or two mechanics still have the right tools and they can fix 06:40

your engine. So why are we considering distance from Plex? Cuz it might be the first time we look at it. It might be multiple layers that we never 06:47 have to touch but you know when you're near can can fix my car engine and why it's about 06:55 knowing what you want. Document and I'm planning out. I'm testing really really touched. That's where you will pull over if you don't do that 07:05 profen. So yeah, so so we had fun with this dude's recently presented at the the Great Plains XL 07:15

event. We did some live polling and it got some good feedbacks least it was tried out. So we have this QR code here which may seem to be alive full 07:25 organ. Ask some questions throughout the presentation. So go ahead in the first question is who are you currently implementing multi-site today? So go 07:31 ahead and scan it vote and please vote. I wasn't look like an asshole 500 push-ups in front of everybody and 10 the last 07:39 day to vote good. All right, two people raised her hand. All right, 07:49

let's move on. Right. What's a good results from rolls? And here we are 07:59 change your answers. 08:06 Wow, so how many votes 08:15 so it's 08:17 rock and roll move on. So why the failures happened to him? So it was talk about why failures happened. So you see here in the red, you know will be 08:27 unplanned outages in the black while plan at outages. So what has happened they will happen within a planned or unplanned at some point. Something is 08:34 going to fail which one of best plan for that and we'll talk about ways you can best plan for that. But there's times were things out of your control 08:43

this happen right natural disasters and so forth. So you see here, you know a power outage. I'm playing storefront goes down, you know crashes on plan 08:50 though. It was a bug so we can plant outage for storefront fixed, but I'll I'm going to cut your internet links, you know, make money conversion to 08:59 the same location. You're building something to cut the wires. So that's unplanned. We have planned for having different pass coming in in different 09:06 locations of our building to make sure that if one wire cut it's not cutting two wires, right and then we look at okay, we we have all these power 09:12

outages are unplanned. We have these issues are unplanned. You know, let's go through an environment redesigned it say, okay. Let's eliminate all 09:20 single point of failure. Let's make sure that we don't have any issues happen to us again. And it's a multi-fight supporting technology. All right. So 09:26 let's talk about the supporting technology for multi-site and Johnny talked about failure has happened got a plan for those failures chili failures 09:35 across an entire stack of Technology Citrix is is is somewhat complicated to get set up and that's how you can get a set of pretty easily from a 09:42

multi-site cinereous all the other components that really need to be in in lockstep. So let's talk about just throw some hands up like it's someone 09:50 named some technology that should be supported multi-site. That's it apart of it the one name something. Ocwen is where I'm at. 09:56 AT&T and things like that, 10:10 but really if you look at each of these categories, this is just at a high-level you have authentication which is huge and authentication is a lot 10:19 more than it used to be just active directory on friend. Now we're going as Rady we're doing Federated identity and things like that all the different 10:26

provisioning Technologies. What are you doing from Network respected? What deployment type are you going to do we going to do on friends are you going 10:32 to do actually don't do hybrid or whatever and Johnny's going to talk to that in a few minutes and point management content collaboration application 10:38 in Dayton workloads is critical. So these are some of the Technologies and why you can look you look at their store, like holy crap. That's it. That's 10:45 a ton of technology and architecture reference for this diagram together. And this is how we're going to kind of 10:51

talk to the presentation go in Ford case mentioned in our business considerations first, so things like Risk, Be there is right 11:01 RPO young recovery Point objective. What point in time do you need to go to in and then he's our business level questions isn't something as an 11:11 architect that you would make that decision someone at the business level would RTO from what what you know, how much time can you afford to be down 11:17 deployment type things like endpoint management content collaboration workloads is absolutely critical because it's not just about your application. 11:23

If you think about your application you break that down there so many other pieces like Steve Cohen falters excetera. Look at that. I'll be across you 11:31 to be Services Chico. We talked to a lot of companies are doing in the past and now we're doing a always-on availability groups for 11:37 then recently basically is unavailable Hebrews came in which provides a lot of value math learning provisioning networking Giants wearing provisioning 11:47 with cases while in active directory, they're often forgot about her things like file Services. I just had a scenario with a customer recently wear 11:57

false River went down to have a secondary name is Faith Ford EFI send as an example, right? The directory printing in in platform is the hypervisor. 12:05 I'm getting the Jesus. Stop talking things are going to go ahead and enjoy it right in and data center design. We're case is not going to drill down. 12:12 I was trying I thought you were going through a thing to a crippled guy and 12:18 expect them to forgot you two broken foot time Dad that I read it I start off with this design. 12:28

Although most techies might think it's it starts off with technology because that's where we feel comfortable. It doesn't start off with the goalie 12:38 starts with business decisions to make sure that that you know your business, you know, your business requirements and from a multi-site perspective 12:46 if we have this is not willing to invest in multi-site or they don't see the obvious reasons to do so and they want to stick to a 12:55 single Data Center and you're stuck with that. You'll have to design around that single day to send it. If a customer is more open to to multiple data 13:05

centers or your management is more open to multiple data centers, obviously a lot more options there. So this is just a flowchart. It's in the deck 13:13 and we'll share that with you. So I don't go through each tapping individually, but it's good to have a workflow there that help you guide through 13:21 the business decisions around multi-site the brightness mainly around how you break your site down. If it's active active active passive active active 13:30 tendon will walk through those again from a business technology perspective. So turn so Citrix deployment types here. So 13:39

we'll talk about different appointment times going forward right now. So on premises is the main one. I'm not going to go through each bullet point 13:49 here. But the main thing is an on-premises. What are you do your primary data centers or secondary data centers and so forth all in-house you have the 13:56 most complete control. All right. That's the main point take away from that is if you have the most control if something goes wrong if things go 14:05 sideways you can easily find that will not easily but you have everything accessible to you to find the root cause analysis to look at logs. And so 14:11

for there is nothing to is that image management. So you can automate that use in-house tools. We may have to change them a retrofret them. If you go 14:21 to the cloud or the hybrid model in the next thing is is going to be yes, you'll have higher operational costs. But again, everything is in house. 14:28 Everything is under control. So when changes happen if you have more control of that when things go sideways you can help find information and so 14:36 forth. Can we go to the McLeod model named you're running everything in the cloud. The cloud is is nice, you know, it might be easier on boarding in a 14:43

quicker time-to-value possibly but you don't think about troubleshooting keypoint there. You may not have access to everything if something goes down 14:51 if things go sideways a good example is something happens and is your like they've happened recently you may not have full visibility into what 14:59 happened what happened. You just go into town and it's causing Interruption to your service into your users. And then look at image management team is 15:07 managed that you may have to retrofit or change your tools to do any kind of automation that you may have had in house and look at cost. You know, it 15:15

may not be cost up front. But one thing at the Cancer with class is that cloud is always on cloud is always charging. So look at things. Are you have 15:24 your things that have to be on mix of the stay on but then use power management capacity management to turn things off when they're not in use when 15:32 you don't need them. The point is you're going to put stuff out in a public Cloud definitely look up your pool business hours in Destiny. Look at the 15:40 white lights, you need to run outside of those housing and everything else that you've done and make it light have more disposable be 15:50

having to 15:59 manage infrastructure having them in a certain things. So the operating cost goes down but doesn't he doesn't mean you're going to save money think 16:09 about if an outage happens, what is that cost to your business when you're going everything in the cloud and you look at something like Sybaris Cloud 16:15 wear at the Hybrid model where we have our infrastructure in house in the cloud and only there were closed on Prim and you don't think about that to 16:21

where you're not having that day to operations of patching your SQL patching your controllers maintaining all the infrastructure. You're more worried 16:30 about the cloud connectors on you know, when your locations and so forth now troubleshooting YouTube Bring some back because you have some things and 16:38 house where you can still troubleshoot find logging look for that root cause analysis but you may not have all the logs available. If something 16:46 happens in the control Point organic something like Azure goes down into having text your environment image management this anymore type closer to 16:52

on-premise is because you can still use the same Tools in house and then cost again will happen. So you calculate and plan for failure. 17:01 MN hardware and core components These are things that might not be under your control. Right if you were the desktop team or the Citrix 17:11 team, you might not have control on Harper, but we'll go over them. Anyway, because I think it's it's important to ask but even if you don't have the 17:21 control go to Your Heart by team go through your networking team go to the data center team if if there's a specific data Synergy team some about 17:30

these things right be aware because we can design around these items but we have to be at work and items here or protect protect against failure to 17:39 have multiple power circuits try to eliminate failure as much as possible go through each of your individual components that you have and try to 17:49 figure out if there a single point of failure or not. And if there is still you see how you can get them in a Tri-State and obviously conductivity and 17:58 this is from outside or from inside outside. So I've seen situations where the business Went down because there isn't a circuit from 18:08

inside out when I went down because of somebody cutting a cable keep that in mind have multiple paths and that goes for the 18:17 services as well services not working. Like you're supporting him from DHCP 18:27 DNS. They might be like you might consume them as utility or as like water or electricity, 18:37 but it doesn't mean that if you feel over to another side the services are there or responding in a similar similar way to keep asking those teams 18:47 and obviously you need to replicate your data across multiple sites as well because the whole core functionality of desktop virtualization or 18:56

application fertilization is bringing apps and data close together and does not make sense. Like Dave said to have data in one region and have your 19:05 apps in the other because you're introducing wait to see their And the third one is think about your core design think about your networking but not 19:12 just on a Nexus fast, but you're not working and maybe your storage networking have multiple that have them 19:20 redundant and the same goes for Power Pad. So have multiple power circuits have your equipment hooked up to a different Power circuits as well even 19:29

multiple power grids to so you don't want to have everything for your data centers in the same power grid or down the street a couple miles away 19:39 because if that area of some natural disaster happens when you lose power for stained amount of time, you know, that's a problem right there. And so 19:45 that's why we look at these cloud data centers Azure or AWS DCP. They've got it totally different regions and toys from Power grids to make sure that 19:52 doesn't happen and even generated can get him to a pharmacy. That's a very good point you making now. Dave about networking. Yeah, that's 20:00

three things on network and he's going to end up in your pot. She's got a bad network if it's a slow Network. 20:09 Hands up who's had the Citrix is broken and then hangs out who has been going to the 20:21 piece. 20:29 He's talking to each other driver wise about traffic manipulation, especially in a Mojito environment. So when it comes into a region, you need to 20:46 manipulate that traffic based on criteria the weather is going to be in a key. 20:55 If we look at Design This Is Fundamental to resilient phone platform 21:05

oil today's 21:09 Network 21:18 segmentation into multiple regions. Those rules need to be defined upfront or you're going to have an absolute now. I'm at the clouds. So 21:30 you'll have your network planned out before 21:38 you start to go down the most so I parked. Resilience is like I said, I was going to be you need to provide multiple puffs. 21:48 They sense that you iron whether you'll come from an on-premise estate center of the cloud even to the point of the the hypervisors that you got in 22:03

a wreck. So I have a couple of course, which is the top have them on separate power rings have a couple of tables in each node, Mason to go to 22:13 District Court switches and provide Brazilians within the DC and then who's hoodkroft DC. I never 22:20 use the network off together and make it more resilient between you and I sent it 22:28 from a multi-site to stuff to go with going to be on. enormously technology there 22:37 is loads and loads vendors out there that offer technology to help you with delivering multi-site Citrus deployments now oversee 22:47

most of the time and you know, I'm a sensitive guy. So most of the rest of this presentation be based around ABCs 22:56 and how you can use from a networking point of view how you can use an ATC to help you with multi Syfy wire using 23:06 traffic management between you and 23:14 you ain't text me about my 23:22 application load balancing failover. 23:26 So if anything just no excuse not to have any kind of traffic Management in your environment. 23:36 Okay. So one of the choices in in in that whole stack as we saw it was was hypervisor for for your multi-site deployments as well. And maybe 23:45

maybe it's something that we have to go through to get you get to understanding why were mentioning this still looking at the I prefer 23:55 the artist formerly known as 10 server. If you work 24:04 was good, and there is nutanix, ahv 24:14 supervisors that are supported on premises and you are in your data center sitter. When When selecting 24:22 hypervisor are quite broad Miss cost. It's a management. Think about some of these hypervisors will have a 24:32 single point of failure in their match itself a couple of these like a SUV or RV Center will be 24:41

your single point of failure in your architecture where Citrix the master in the pool, if that goes down if there's an 24:51 election and another massive becomes another member of the pool becomes the master till you were infrastructure management is always there you were 25:01 always be able to Can influence the the rest of the intersection obviously adoption is the key item look at Community Support. Look at 25:10 people around you that are adopting these hypervisors talk to them. Make sure you're comfortable with their performance. Obviously, we were for 14 25:20

days does Lodge around performance superformance is one of the items that we complete touch upon when we look at hyper versus selection and 25:30 specifically more to do this session availability echo system and features that are important for your multi-site design. So is your your hypervisor 25:39 able to expand across multiple sites is your management playing able to expand across multiple sites. What's the echo system looks like do you need 25:49 tools or zurdo to do data replication from your I provided to a secondary site for new tenants? 25:57

And also looking at the future sets like having a fan clusters or if you're looking at gpus think about that as well. So 26:10 selecting your eye provisor during a multi-site design isn't is one of the key features here and obviously update Cycles would apply here as well 26:20 because me with with all the stuff that Intel's been pulling off lately. You might want to consider something. I didn't update it in regular basis and 26:29 as though those patches across those multiple sites. All right. Talk 26:36

to us briefly about screws and delegations. We can pick things up and you actually get into the music presentation. I just want back what's go forward 26:46 to basically three key areas here just gross delegation and writes all really around. The area is usually what and when I seen it a couple times going 26:53 into customer environments. Are you go in and informed architectural perspective you going either you hire a consultant are you doing internally build 27:01 his beautiful design for for your first sight? I'll send the business comes down the line like a we got to expand we can either move in Highbridge the 27:07

cloud or getting a secondary data center and some of those initial concept that you were so stoked on for design like security rights and obligation 27:15 or forgotten about so you want to make sure you prevent organic Bros like that and take the time to look at the business considerations will get 27:21 proper roles assignments excetera, but not just across Citrix right in and catalogs and delegation. Are you want to look at all the other areas like 27:27 replication your profile storefront ATC or you GPS if your policies all those areas in the staff you want to look at and really just keep that Focus 27:33

around least privilege in in delegation. That should be your number one focus when you're given rights out in general and just want to mention real 27:41 quick here is ISO delegation and configuration logging was added fairly recently in in in Citrix Cloud was something I was missing for a while. So 27:48 right now that you have delegated right single user today, it's not group base, but how fast are generating? It's only a matter of time before that. 27:56 It's so just FYI. Skip coming for the right on target, 28:03

but it's very important. And in fact, we actually put this in here for a failure to Maine for the failure to 28:12 me cuz actually I think when you guys supposed to talk about failure to means but now we have this in here is a failure to mean for that topic to do 28:22 they did their value it against rest, right? So what do you consider from a risk perspective 28:27 give you an example right it from up from a site designs. Tampa guests were talking about you might have taken affect appointment right in and maybe 28:37

you can do it in a have two physical sites in any other of their Judge Judy located very closely dark fiber excetera young for that and a prospective 28:44 make sense into a single site. Do you maybe do multi-zone and then split sequel cross tattoo, we're good to go. But from a fault domain perspective 28:52 when you go to do an update there is risk involved right if you updated as a failure the Davis you take the whole site down, right? So that's how I 28:59 would see what that they recommend Saint Anejo separate site or if you're doing you know, Turn on failure domains for if you have that scenario 29:05

basically where you want to have to say 15000 users, you know, what 50,000 users on a single site you want to split those out so I can pod based 29:12 architecture. We have your workloads, maybe a control services and SQL in separate piles of 5000. For instance. If for example a rack to be a failure 29:18 to be a monkey on the side 29:25 there and we have an idea. Your business Monkey Business 29:35 have 29:42 you asked me so we never let this happen again? They went and built some some really hardcore scripts to proactively go through all aspects of their 29:52

infrastructure. It started as just a chaos monkey in adult a hole with a call to chaos Army. But If You Lie part 2 the infrastructure during the day 29:59 and they basically from that they learn how to have proactively constantly generate their infrastructure data plan against failure. So if you're going 30:06 to if I fail you want to feel fasting and learn from that, I'm just pissed about those Florida man. Just going back to that point obviously technology 30:13 is evolving and it's allowing us to grow bigger faster. Might not make sense to look at the biggest 30:22

pods. You can get depending on your retirement. You might want to slim that down and we've got examples for people were using PODS of 30:32 32 nodes are 32 servers going back because their maintenance window was growing out of control or they couldn't keep their maintenance within that 30:42 maintenance window. So that power and is pushing for my CPU Ram in this perspective. We're getting more and more power. We can cram more 30:51 more users on a box can keep an eye on your Filly domains when you're deploying down and sizing this out 31:01

fertilization management structure like the center and things like that, right? This can be single point of failure. So I'm so Tiger 31:09 stripe design plan on the deck certain color. So I threw the donut slide in here because I am sorry. I think it's funny because I really enjoy 31:18 Gourmet donuts. And I think that we know your workload right you could ear ear is really like at a box of donuts in the storm are donuts near your 31:28 work looks of all different characteristics and my favorite is the red velvet donut. If you look at that from our lairs, you can say like, you know, 31:36

the the the the cream cheese is a presentation layer and maybe the ingredients are the what's what do you 31:43 mean by now you work with I wouldn't I was totally not going to go through a stack of donuts. But but it keeps saying here is your workload is a split 31:53 across multiple sites and how your work go to swim across sites. Things are coming to a play there. Are you going to have localized applications? You 32:01 might have Legacy applications on site your application might have identity that's Federated through either OCTA as read Repaying excetera other 32:10

services your presentation layer Citrix. This is what we're talking about the workload. So critical you have sauce applications of different types of 32:18 Oz movie up infrastructures workload applications Mill Center stock is the things that you have to think about when you're talkin about your workload, 32:25 right and that's going to determine if you're active activewear active-active 10 right where you're basically both dated centers are live and active 32:32 but you're pinning certain users of the Datacenter where it makes sense, right? So things going to look at it see on your apps. Are you up to your 32:38

file Services SQL Services Network from a from a ports and protocols perspective whitelist Blacklist that sort of stuff. These are all things. You got 32:44 to think about identity as well. So very important take the time to understand your work clothes when you take that once you understand you work while 32:51 you want to look at your site is on, you know, you going to do a single site. I can do multi-site razones. How does that all breakout swing through 32:57 this will flow chart together and we made it super small. She couldn't read in the Box I can say anything you want. I want Now does not but basically 33:03

it's a couple boxes because we want to move on to get to the devil's in the fun stuff, but we wanted to break We talked for a long time is it what are 33:10 key questions? I can help us make the design decision. So Princeton or do you have more than one DC? Yes or no. Do you need internal? Tell your 33:18 domains? And then the last question once you go to that is basically do you need to know the way management costs and resilience and based on that? 33:24 You're going to help determine what that layout looks like and design process really goes away when the multiple sites and are you and see 33:31

classpect is gone right. Now, you're just talking about multiple physical science, but those are worth really resource locations. Which of those right 33:41 there we go to the store really quickly here. So the first one in the last time we threw some 33:48 Geographic stuff on a compass and perspective but single site multi-zone Miami New York l a different zones sequel can either be in Miami or you might 33:57 split sequel again depends on the wait-and-see on on how you do that in and we'll talk to see KO on a second. So I want to steal from Thunder from 34:05

there a single DC multi-site. That's if you need that internal failure domain would look like what's Hello, you have free physical science, but your 34:12 access layer which day was going to drive through. So I won't steal the Thunder there you're going to see things like resources and stuff like that to 34:19 have you don't have multiple icons and stuff like that a multi-site primary zone. So that's basically no single site to separate physical sites again 34:25 to an aggregation and you can actually do multi take multi. Tell him where that Hub spoke design, which is USA nice and big there and then you have 34:34

the other site. That's that's over in a Mia different different zones are what we had a discussion about that because I put it in Europe from the UK 34:44 TimeWise 34:54 on basic always-on availability groups or something that was fairly recently introduced in the sea. Clothing was a signal 2016 people standard, right? 35:06 So basically main difference between that and I was on availability groups. Basically going to go do the same fundamental infrastructure as far as 35:14

requiring sequestering you have it but the different the main difference there is your rear axle listener right have a single listener for database 35:20 vs. Where always-on availability group you can have one listener and multiple databases and failed them all over and sets with basically is on each 35:28 one is kind of handle differently. So we're in the key thing there that that also was a development as as time progressed was you had to have a foster 35:34 witness previously. So you have that third data center in a multi subnet design Which Wich in if you didn't have that then you can have the challenges 35:42

with Purim to get the file-share witness. They read the Recently Added as her as a witness as well. So it's basically was just kind of needed a little 35:49 storage Bob there in an axe is a third witness. You don't need that third Datacenter, but the key thing on this slide to talk about hear that.. I've 35:56 seen a lot of folks mess when we going to do a lot of help taxes. If you don't set the multi subnet failover string on the connection spring on those 36:03 databases, you're going to have challenges when you go to failover because you think about it when you do when you have multiple subnets on the list, 36:11

Those are two separate. I see dresses that are registering an Indian ass in a traditional application that doesn't understand how to handle most I 36:17 said that I didn't respond back with different IP addresses. So this string allows the application understand on which one's actually active. So be 36:24 sure to set that and we'll talk about the other reason why you would want to set the why you would want to start actually know what we said that for 36:30 us to look at who is Cash mode. If you going to hook hose cast non-provisional pulled machines on how to operate correctly there would look with those 36:39

kind of things. Should I move we move on when you in the face if I have one of your 36:47 one of your nose is in your make sure you had the IEP students faces for your back while I try 36:56 pays in Israel. Good wallets Windows clustering look like it. Come on. You want me to get that and we had this in love you trying to get to it in the 37:06 three-point and we actually walk through that here in the in the video. So we're going to 37:14 hit her first video. It's it's 20 minutes long and we're going to walk through it over the table fell over demo courtesy of the other cool bumper 37:24

hear a case. Very cool. Seven seconds allows me to do my moves design. So basically have it split between 37:34 UK web booking Azure UK South. We have two nose in her failover clustering you can see each of the bags that are actually a part of the sorry. It's so 37:42 blurry. I want to make sure I would need glasses after this I can see your Cloud witness that's over here cooking. 37:51 Basic always-on availability groups on SQL at 1 you can basically see that we have inside of these database or a Tissot always-on availability group. 38:08

We have the listener listener has IP address on each subnet, right? We have the port 1433. I'm missing you notice and will register. Her name is 38:15 becomes names and DNS and that way if that's how it's able to register for navigation perspective in the database yet either synchronous or 38:23 asynchronous. This is important. Now that requires the right to happen on the other side and acknowledge it. So if your split between two physical 38:30 locations to do synchronous Raw tonight fill over you're going to see that the consul's actually pretty slow right depending on where you're at. We're 38:36

fine where he is or when it feels over town just kind of showing their that that's their split between the two sides in the back of a second and this 38:42 is important. What day was talking about it sequel 02. That's an Azure ratchet showing on networking and on the actual physical interface and Azure 38:48 you can actually see the IP configuration. Do you have to go in to find those additional eye peas for each of the basic on availability group for 38:54 listeners as well as the cluster ID if that's not that it's going to break so Play just a little bit here. But this is this website of this about 39:00

TTX article. I talked about this is talking through a set of strips for changing your connection strings in the in the case of failure or where you 39:10 need to set in this case the most I said that availability trip. He basically it's just a Powershell script to kick it off. It goes to all the 39:16 controller's a numerate Seminole actually go ahead and set that string to True vs. False and then I pressed it and make sure it's ready and available 39:23 for for multi-site Designing and Skip ahead a little bit. So it's go ahead and make it fail rights are going to skip ahead and super fast computer. 39:29

Dave's got in the lab and my right here we go. So I seek you out one going to going to prison and go to sleep in a showdown rights equal one just took 39:37 a dive dead it going to Cluster fell over and take a look at the the actual nodes and we can see that SQL to SQL one is down down hard users are 39:45 pissed off she goes down but I don't know about right why don't they know about it? Because the bags filled over seamlessly there synchronously, 39:54 right? / 2 / 2 to Azure. So let's go ahead and pick up Studio Studio by Design is going to snow. All the clothes starting up. Also Dave like a signed 40:01

one gig of RAM to hear. What I want to 40:10 show here is just in sequel to you can see when we kick that open. You're going to see your your basic on it. A Day always on availability groups are 40:20 now set to primary and everything filled over successfully. So it's pretty awesome. So just wanted to show that and how is split between on Prime and 40:26 I sure we cover that up pretty well there. So look out for Pascal from licensing design standpoint 40:33 overseen for Citrix work you've got a decision to my 40:37

other point, whether you're going to use it to use traditional on premises licensing 40:47 issue the same license twice to two different service and have a new life 40:53 unless they're the same night which cause an issue if they're on the tonight with some people but 41:03 we don't get into that sell you have oceans around all the clustering or light balancing. Load balancing you can have two machines at sign name off 41:13 the mine in walk two groups an issue with licenses ones put it on by things and then run active passive with a citrus IDC in front of it. Or you can 41:21

have a Windows cluster an issue the license failed to the cost to nine, which will then put on bites not itch and then you can use when discussing 41:30 Skylight make sure you replace certificates on the license in one place 42 for the web services and the licensing Services is 2 41:35 places you need to replace it. The reason you do that is so that when there's a fail and when it goes from one side to another you might need to go 41:45 into oil services and get Citrus Studio to retrust devices. Yes. I trusted you won't get that 41:53

Services. 7279 Singapore 42:01 goes down a DNS authoritative DNS Services. Make sure that each shell on if you're using gs gslb 42:11 on the DNS record. 42:19 Sorry, that was a fighter recently where we had the licenses issued on active note. It was running in active passive. 42:32 The student has been replaced. Everything was good. We didn't have the license is on the second United shame in that in the event of a fight at it 42:41 would drop by as soon as they 42:48 have 42:53 the same size and I'm off to mine in a group with the losses on both because it's as if I'll ever since you found out about it and it won't trigger 43:06

the garage. With the customer International we did have it 43:14 and we'll get them to you somehow. This is the way 43:21 we would typically set up a dentist. 43:31 Since you said you see sitting up front 43:34 is coming on monitoring license ever having lots of passive one being on premises and the other one being in a juul. and if you lose one if I was I 43:43 was so loud. Or that problem goes away if you use Citrus County. It's a 43:52 Service that you access and is managed from your city Cloud control plane is actually says you can see that 44:02

you pull in a built-in. Can you get that noise pretty grow out of the box and it suppose you 44:11 drive so you can as you get more years you can just have them up to it. And that's it. That's Larsen C moving on access layer. 44:21 first one fire hose versus 44:32 and which ones the direction you got 44:36 head will change the way you look at how you deploy this in the most so environment? What a storefront it is according station and project 44:46 has access to your virtual apps and desktops remotely. Is one of the single most important things you'll going to manage 44:55

in your environment because he's what you use it tops every single time. I can see a massage from you if you plug in a cell that going to moan a lot. 45:05 So I need to be in this real quick. You need to be able to move around your diet centers in the uses of need to be at to get to it. Some of the things 45:13 you need to consider in a multi-site in a Mozart deployment of storefront considerations about a subscription. 45:22 A subscription stool is yellow gone you get app that shows you click on the favors and I got you five hours link manager still there will just that if 45:34

you have a subscription stool and you have set for a Chevy Cruze fuel and service in the different regions. If you get fired 45:43 and you haven't even know if I'll ever see that none of their applications will be displayed. There's an art club 45:53 that where you can use Powershell to synchronize a subscription. Do I switch between storefront? Have a great night and see if she 46:03 can run Or the locations you have in the subscription database who does synchronization will be taking the 46:13

cat for you. Yes. I love you from Tyson. Tri-State sense be and then 46:23 everyone there and you're still sinking only that one way and it would like to write to me what it was hitting. You need to consider his car as soon 46:33 as I'm still multiple versus single Chevy Cruze. You can have one that's the other thing you can do is you can have two stores on one still 46:40 friends forever and pulling them both to the same subscription-based way. There's no way to tell you how to do that. 46:50

If you have multiple sets of delivery groups attached to the storefront service. Then you're going to get 46:58 what you sell from screen that you're going to get to two iPhones to run the same application that use it on her like that. So resource aggregation. 47:08 you can add you aggravate you control groups in that I saw uses to that and then they got a single lock on the stove from will handle West run that 47:17 from Ogio works, like when I use of hits a Gateway 47:24 when I click an icon storefront will give it back the location of the netscaler gateway gateway in The Diet Center closest to the 47:33

application. You can have multiple Ogio. So we have one hit for lip hook and wonderful as you will you pass out when you click on the link and it was 47:43 in the use of direct vacations Simmons, you get the microphone thing something to consider. It is just I'll be in front of you'll often 47:51 can I buy Ridge so that if I didn't go to the other one performance went quite busy is good, but you don't need to 48:01 switch things off refresh the page see what 48:10 happens to other behaviors like Local site like better since I wouldn't date since load balancing ngsl be across and the reason why density is out 48:20

there so using ifs's example, if you have it sent right and you have a Hollywood is so fun in Tyson Rent-A-Center pay, but you don't do the same as I 48:29 hear fashion. You lose this song. I'd still friends cry if I was over everyone can get as soon as I touch One log on because the identity platform 48:39 isn't made multi saw it wet underneath you won't get a logo onto applications. So 48:47 this is a storefront Mi which on I'm going to skip cuz we 48:56 finally is where we fight between on-premise in Israel. 49:01

cloud-hosted middle of back stretch. Alright, everybody get up and do the YMCA do some backflips. Do you want from there we go. 49:12 Citrus West face is a service host you. Have 49:32 you have? Applications that you can host within your date sensor web app so you can 49:40 host in your Dyson and you can have on the premises about traps in desktop. Send. And Citrus county 49:50 is going to use 49:59 all the people who makes 50:02 it is a service provided by this situation 50:12 never really say that is 50:15

because you don't need to worry about Maltese or delivery of situation by situation worried about that for you. Will you do need to worry about 50:23 it into jewelry Source locations from the situation. That's something you need to worry about such worry about making this service available 50:33 all the time. When is you know, if you're if you're doing if you don't have localized failure and desktops need to have a storefront netscaler on 50:43 front workspaces. Additions to up to her and everything else today announcing the kenite body weight rule apply to 50:52

service and I did get with relation to that is enough to be spoken about is heavy heavy on Brandon's girlfriend email get what 51:02 you want from Citrix workspace and there is so many morons Creations on this is how much for security and this is right store for non-steroid 51:12 option moving on remote access the cloud on 51:20 premises. You have a situation UniFi Gateway runs lady. See you can run it on as usual. I was gcp physical 51:30 within your tight sensor. Consider the difference when your deployment as a difference between the differences between citric acid 51:38

from about four regions. Make sure you consider that ticketing is something 51:47 you need to consider because if you have tickets from a single Diet Center and you fight I will. 51:57 I open station the same applies for UniFi Gateway off as it does. You need to have that available in multiple regions if you're going to fail this out 52:11 of it? Things in order to have you all right. That was to make sure cuz you can do pretty much can a miniseries independently 52:19 make sure it looks the same as something really simple. But if you get or if you're ever use it working on this Sunday FiOS Internet coming the next 52:28

time you go out why the default black guy that played 52:37 Cloud hosted 52:40 Citrus Gateway Services Hollywood multi cloud service that again so you don't need to worry about how to make this 52:49 multi so I know where they've done it for you. Asus em5000s offered by Ascension cloud 52:58 There's nothing can figure out its own by difo. When you sign up stitches clown is going 12 County. I think maybe that would change the 13 53:09 Global points presents. Not so much if it was a it was 12:00. You can configure and truck stops in and assess have to run through the centuries 53:19

guys headed. Important for you know, 53:29 what goes down when it goes down and so forth either try to be proactive or reactive very quickly. So the first thing we talked about is Citrix 53:39 director, I'm not going to go through all this but the main thing here is that if including it's free based on your entitlement you may have longer 53:47 retention Langster retention links mail application pro pro me when she helped you kind of stay ahead of the curve kind of see in applications are 53:53

available or having problems. The main thing I want to say here is that if you're doing load balance there multiple direct the servers make sure your 54:01 customizations are the point. I'm all good Rector nodes also too if you're doing any custom filters, make sure you copy those across all your nose and 54:07 I won't be there or Point them in the same site to a central location and so forth. When you configure director in the discovery address, you can 54:15 point that to a dsld address by default. He wants to 2.2 One controller per site on the nice thing is I was at want to talk to that controllers. 54:23

Cashing in a second-floor goes down. It will still function in a diversion from intestinal we've been doing but again you can also points to a dsld 54:32 address and then for another site do a, and then do another DS all the address to worry about it doing that cash your head that automatic or have that 54:40 field of work. You can do them to Twyla gslb. Too quickly touching on ATM and how this would work in a in a Mozart influence. 54:47 This is application delivery management to monitor things 54:56

across him always hard time full video 55:06 control free sample SQL you want to know if she was walking because you're relying on it, but you might not have the access to monitor setup. 55:15 Any event that NASCAR to channel 286 generates he's going to go to the idea. So then create a rule on the idea that says if you pick out this 55:28 way of using this to Monica things in multiple regions, the only thing so you'll probably need to bypass 55:36 one of these days unless you have a message on Pecos buy stuff for nothing from one of these. 55:46

So use a Persona user Persona is important because if you're in yours or don't have their stuff they're going to call and do you help desk generating 55:59 events ticket event since 04. So the first thing we talked about his work space environment where to go to this fairly quickly here. So work space 56:07 environment management has kind of inherently multi-site cable with built-in based on how the infrastructure brokers work. So you can still have a SQL 56:14 database basicos en always-on availability groups does have a localized cash flow from a multi-site standpoint at the house in the case that you have 56:22

a Bill Billy issue. So have a localized Castile work there from a from a DSL be perspective were going to look at now and how we can use DSL be an 56:29 active passive in these data center separate broker load balance URLs, basically or hosts names and then do fail over there on the other side to 56:37 basically make you always on availability essentially always on their keeping me going to look at here is basically designing your Euro you structure 56:45 in your in a configuration sets accordingly. So if you want to have a good OU structure built out you have that all used trucks. . Inside of a inside 56:53

of wind and then what we talked about different hosts names for each of that each of the sites. You identify the GPO set that broker a broker 57:00 hostname. Basically for that Jesus. I'll be around a specific site. You always Roberto. Localized brokering that's happening. If there's an issue than 57:07 it was broke her that way, right? So that's kind of cool. You could do that. And you consider looking at using I-4 upm. You could possibly use upm you 57:14 can use upm so that as well as recently I mean, you're not a hundred percent in lockstep to wear like the upm releases are so if you want to be on the 57:21

latest version of you p.m. Still consider doing you PM management outside a whim but it all depends on your specific scenario. So I'll take a look at 57:30 one thing that that was kind of need that we did on a recent appointment is using location variable pass so friends then saw this is where you start 57:38 using the technology it's built into when they kind of do some new stuff for multi side. So if you take a look at the two data centers of CTX data for 57:45 the DFS namespace and we have 2 2 2 2 2 2 location to the printer variable price of a print server is another thing for Besides, right. So what we did 57:51

is CTX data for the variable if we look at the actual variable name is ETS date a live talk-space see the actual name space for DFS n rights. We have 58:01 2 DFS then namespaces for DFS the print server. You can see the name there. So print server is a variable but we have the actual after UK print server 58:08 for the key. Here is the conditional access right or the condition for the filter where we doing it by IP or something at base that way if we log on 58:16 to have UTI over the secondary side UK South it is said it does it when runzo I'm part of this go out of map to cry print server in Quran about the 58:23

correct profiles. If you take a look at the actual action for the printer you can see last last print server. There is no 58:29 title to this someone point out shines glaring error on that side. Yep, the phone number for the wrong 58:39 section number on the office is you have your filter for the website as well. As I your 58:49 environment variables are going to be attached from an action perspective within you have your filters that are assigned to that so that way and then 58:59

you never give us news faces. As I said Miss basically allow you to have you got dfsr from the replication perspective but you the key when you doing 59:05 give us our for profiles Thursday me on my cell phone support. I think he has you have a single Target. Are you going to lend you have to pin them to 59:13 those dates everyday that centers you can't be active back of the same time. You can have replication with you and make sure you're on a single Park 59:18 in this is how you can make sure that that happens with the with the variables are so I Feel Love or design or just go to this guy here so base. We 59:23

didn't I already talked to her pretty much but just doing the load balancing GS. I'll be having separate hostname set up for each site. And then do 59:29 you act the passage? Yes, I'll be on that host name and then separate GTOs which which will look at actually in a few minutes. We talked about video 59:35 design user Persona management. So you in a couple options here first built. Microsoft but honestly after Microsoft has purchase fslogix, we shouldn't 59:42 be seen roaming profiles as much anymore in the field. You know, everyone's looking at this Logics or you know, what district you are entitled to a 59:50

upm. Let me look at the Hydra solution with with you p.m. A combination of files in Vista X-Files and then contain fully container-based things. Like 59:55 he's really risen fslogix the main thing though when you get to these vhdx containers as they're always being read and written to you to think about 1:00:04 how do you make those with redundant? NF logic has a future call Cloud cash, but then things like user layers or the Vichy accident and you p.m. You 1:00:12 can't just copy those anytime because they're being written to all the time to think about other ways to handle that I think about when you want to 1:00:19

use it for Saint options, everyone wants active active, but honestly active-active premature pipedream uses more of active active pin or active 1:00:27 passive how much you have ever use on your environment to maintain profile fit active is real or if you're doing any kind of user Persona 1:00:33 active-passive across multiple sites in the pipe to be about 3 with that. 32 pipe dream. So but you know any solution were titrated and so 1:00:39 forth your last right when you do have some application methods built-in. Like I said, the cloud cache has theirs we can use third-party products 1:00:49

under the key thing here. Is that the one thing I've been testing software call Pierre software replication in so doing things like user layers or 1:00:56 even if it's Logics or anything that has vhdx containers built into it don't replicate on the Fly and only do the Delta changes across and a active 1:01:06 fashion. So you can get there appears that they can do source and Target the same solution mix Solutions and and so forth 1:01:14 and it supports all those I wasn't to go through and do a demo by the most of time will post it again and show you actually an active active with your 1:01:23

software of user layers. I mean if I'm going across to data centers with Albia North Star demo that wall will post it. When do ptas next cool. That's 1:01:32 really your work clothes, right? This is important. Call centers Cloud control planes Hamlet there a VJs. What's what's in in your data center or in 1:01:42 your resource location? I just was the meteor that isn't made of what you own. This is your work clothes drafts all that fun stuff. But there's more 1:01:51 than that, right. We're going to talk about his VA registration how to protect against a controller inside failures. I want to talk about VA design 1:01:56

from catalogs Olivia Group after oops a multi-site Concepts. And then when I look at the VA policy so policy is not so much for just Citrix organ look 1:02:03 at active directory or going to get GTOs and just helping out and 1:02:12 how we broke things down. So that looks like from the registration failure Concepts couple of things and how do you register a protein? What are the 1:02:22 option for registering a vdi? Right? So you have a local policy right to local GPO base policy or you have a GPO that no actual 1:02:29

active directory-based epos a local or or active directory at the has the highest priority registry base that I could be a conditional access. Face, 1:02:39 if you have we actually did this Riesling for a customer over my grave in the Citrix Cloud the condition pee so we did that the list of PVCs and we 1:02:46 did a condition based on the computer name of the actual pdas. We had in percentages of the bda's fully register with with Citrix dog connectors never 1:02:53 say that, you know, when to separate separate Imaging catalog and that sort of things you can do that there as well Legacy is the old lady message and 1:03:02

then yes, I really do the personality to i&i most folks are doing most folks don't realize what they have auto update enabled and I just kind of want 1:03:08 to go to install the VTA next next next to put the video in and they're actually using auto update which will talk about with install next next next 1:03:17 station, 1:03:23 right the registration concept. So there's a lot of concerts. What were the ones that specifically affect multi-site is the auto update controller 1:03:33

settings. So that's basically what happens there is when you actually put in your Chores, it's actually talking to the primary controller in the 1:03:41 primary Zone at that point. It's actually going in and cashing in a local file for the name of the file back to have this did of each of the each of 1:03:48 the controllers. Everything is Cash local weather is is control that goes down or new controllers rather say technically it's auto updating. Right so 1:03:55 there's some value there but also that can have some downside in cases where you want to do things like controller grouping we want to do in a 1:04:02

localized VA registration for you. Maybe want to say this particular don't want to make sure all my BJ's in his own go to these two videos and then 1:04:08 maybe you know it in the case of failure that third I'll have you know, you put brackets around those in the in the key of a third controller that 1:04:15 would go to the primary zone or what not. If you're doing controller drooping and you and you end up doing that key thing, there is none disable auto 1:04:22 update setting because I'll actually interfere in and it will be to take precedence over what you doing. I can't are drooping. I'm always on brokering 1:04:29

kind of cool concept that was pioneered by how I don't know if he's in the room anywhere, but but There's a lot of 1:04:36 that on CBC in discussions. But basically they will talk to the failure Center has but you look on the right super simple diagram of essential you 1:04:46 have single site in this scenario. You have two zones. I'm instead of having in that donate friend since I having two controllers in zone a and two 1:04:55 controllers and zombie have three and you basically have two controllers and donate in that third controller for zone is actually going into him being 1:05:02

at Swap. So in the case of an outage with with the primary Zone the studio won't launch can't do any management AAS in the primaries owns down. That's 1:05:08 the problem. You can technically run a command which will then change that over to the secondary and then satellite Zone and you can actually wants to 1:05:17 consult that point. But this was this method is seamless it's called an empty but it was Pioneers empty primary root and it's pretty needed and by 1:05:23 doing this year. You said Sensi up two controllers and infrastructure a primary and then you have your satellite has free to those sites, which is 1:05:30

kind of cool. So I gives you a kind of always on broke rank in the media is technically they're not going to ever register with another controller in 1:05:36 another satellite Zone. Beowulf my BJ's if my local controllers go down because someone forgot to put inside finity rules on the on the hosting 1:05:44 infrastructure or whatever, you know, it'll just registered with another another site and it doesn't work like that timer is on so it's important 1:05:51 or 1:05:58 you had basically what we did actually so if we had we had basically two sites and we had we had a catalog set up and then we basically an Emmy for 1:06:10

general purposes. We did it with no static got to be honest. But if you're doing machine creation Services these guys going to talk to me what it 1:06:18 looks like from an image management perspective how you replicate that image of how the two separate catalogs have your master is over he'll talk to 1:06:24 you that but she said she can have a single delivery group and then though it's going to contain resources from those two catalogs and then you're 1:06:29 able to have in that case sore that that active active active active pin scenario, which will talk about what that looks like here in a second. So if 1:06:36

you look at that from a delivery group stamp. Aggregating resources and this is where you get to pinning only say that what we're talking about a 1:06:44 spinning users or groups to specific zones right as their Primary Care applications, right? So that's I think I have I didn't 1:06:50 Circle there. So yeah, so what we're showing in this slot in this screenshot as we can always go to UK Southwest or 1:07:00 specific applications. Maybe I'd your whole app stack. You have no 5 apps that just aren't active active can't be running. It has a database that 1:07:10

can't you don't have writing to place whatever so base in that scenario based dependents can an application that specific data center in which fell 1:07:17 over after your infrastructure pills over right? Then they'll tell over so you can do you can do restriction C. Can I say you 1:07:24 only certain apps run on these specific controller is not to do it for if you just do an active passive site you can use tags for that is what I need 1:07:34 things you can do with tags same thing with that proved. So I'm getting to look for any Oscars. Will 1:07:41

you design a crappy policy 1:07:45 design and it just makes her a complete Master multi-site you look at that kind of structure like this ever since exciting and production development. 1:07:55 So you want to test I break Under Siege excites. You want to break your your sights down and under each site. You want to have something for like 1:08:02 catalogs and put all your catalogs in there apps to desktop catalogs have your infrastructure and then get then go into the next site and hydro You is 1:08:08

kind of broken. I like that cuz then the next step is you can apply policies and you can have policies so things that we were talking about before 1:08:15 like vehicle registration near where I'm supposed to be registration to the broker. You can do those specific to those sites and different policies 1:08:20 like that. Have a nice hierarchy set up there. So looking at GFS we talked about that before but that's what I'm saying. Here is a referral status is 1:08:27 key is if you only have a single single single referral actually been able to have time. Otherwise you have an issue. So, you know, the date isn't 1:08:34

thank you how to make sure you're always pointing local we're able to do that without Patient variable that we talked about before we are looking at 1:08:42 doing some testing instead of using DFS. Is it alot gif as a Reliance on a d and things are if you don't have two names face overs, that's absolutely 1:08:48 critical. You're not here. You'll take down your whole environment at the hop in that on Monday night actually with a customer. So we're looking at 1:08:56 the possibility of doing gslb and said for the naming but there's some things are considerations to that. We're looking at with the Microsoft stuff 1:09:02

with with Street naming and NSP. And so I got to see if that's a possibility is that that we feel to be much more nice design. We can do some cool 1:09:08 monitors and stuff of that. So we'll see. They are so thanks for ending it instead. I forgot you 1:09:16 were looking at this from from hypervisors first perspective looking at the office that we have. 1:09:26 We are waiting to meet your book is here or non-persistent either machine creation services or Professional Services persistent can be full clones and 1:09:35

an apple earring is obviously something that is is getting more and more attention in in the multi-site deployment. So looking at non-persistent 1:09:45 desktop. How would you go about replicating back from the storage perspective if you look at any major storage 1:09:55 provider that's out there. They'll have a concept for you can have you were single a major single image management inside a 1:10:04 New York and have a concept of storage replication. And that store supplication will allow you to Mike grazier 1:10:14

Golden image to. Secondary site. So that way you don't have to do well obviously I'd say that automation is key to build out your image 1:10:23 Management in in an automated way, but this gives you the opportunity to leverage storage needed needed for application and get you a golden image 1:10:33 from from one site to another. Now if we're doing full clones, it's quite similar, but we have a larger group of p.m. To will use storage replication 1:10:42 and do grouping of a VM or multiple p.m. To replicate that from one side to another multiple 1:10:52

conditions can be said synchronous replication and asynchronous replication. Maybe there are some as soon as you have with the business. I've seen 1:11:01 that in multiple locations where c-level desktops are considered as more important than any other so they will have gold state as they get synchronous 1:11:11 replication. But People Like Us Engineers will get the silver silver model where the desktop is only replicated every 4 hours or maybe 8 hours 1:11:20 again, talk to your business about 2 to them about their slaves and their expectations. So we are we recorded a 1:11:30

little demo here. And obviously I need to talk for 7 Seconds through that bumper to bumper. So 1:11:40 what we're seeing here is an environment where we have AVM it's running on ESX and I've got another 1:11:50 environment running age Botanica town hypervisor. And what we're seeing here is a concept of storage replication. So we've got a remote site. 1:11:59 We've got two different clusters configured here and we've got a protection to mean configured. What we're seeing is that that BM is being replicated 1:12:09

from me is sex to a separate cluster on age. Obviously that brings a little bit of a complexity because we're doing cross hypervisor 1:12:18 migration Norcross hypervisor Dr. But it does give like mechanics in this specific case gives us the opportunity to promote that BM 1:12:28 on Asana said it said replication schedule and add that BM available on the other side and what we're seeing here is I'm forking off the 1:12:37 Clone into a new VM on the HP environment. So they started off with my Windows 10 golden image on esxi configured protection domain to have 1:12:46

first Native replication and I'm replicating. Golden image into a different cluster in a different site. And in this case, I'm running it on age beat. 1:12:56 I'm calling it off turn. VM on in a few seconds when when the cloning process is done and obviously nothing goes wrong 1:13:05 because the VM will start it's things like this in this is nutanix. This is a mechanic no record a demo but old your storage 1:13:15 when there is should have storage replication build in just take a look at the ad if you want to do image replication make sure that it's one of the 1:13:25

technologies that you can use. And in this case. This is my my clone VM. I'll turn it on and you see that that the Rambo starred in 1:13:34 Windows 10 Windows 10 desktop that we can pour coffee into a snapshot and create machine catalogs on on this new cluster still 1:13:43 think about these replication Technologies and think what's possible from from a technology perspective also from a Storage 1:13:52 storage replication perspective. It doesn't have to be bully application layer But Eustace wherever possible work makes sense if 1:14:02

it's available from the application layer like your ddc's like storefront with the global said Lobel of saying that that's perfect. But up until now 1:14:12 there's no way to I to do any triplication without going to out to your store to level so that we go we moved from from ESX on one cluster 1:14:20 to mechanic HV another cluster with our golden image. Melrose hotels, it's awesomeness 1:14:30 us an example here. 1:14:39 We have two sites New York and an LA. We've got PBS clusters in both locations. We got for PBS or be at 4 PBS service needs location 1:14:49

what happens if I got a new image? I've got a new vhdx that I wanted this route across the environment this configuration 1:14:57 right here means that I need copies seven seven copies local am one across the wire is it is time consuming. 1:15:07 Obviously, we've got tools that that can help us there. I've seen dfsr robocopy does the replicator 1:15:17 exactly so there are multiple options to to help us there. But still they require manual configuration and thinks that we have Consider in terms 1:15:26

of bandwidth. And again, it's Error checking as well. Especially if you go across multiple sites and multiple PBS clusters. It can be 1:15:36 cumbersome to do something like that know, what if we would use something like a file server solution in this case, we receive file server is 1:15:46 presenting an SMB share to the PBS servers. We can have that one building in May 4th at 1 vhdx sitting on a file share and when an update comes we'll 1:15:55 just to have a copy across the wire Omar PBS servers are able to reprieve from that new golden image and we're done with your benefit here is that we 1:16:05

can utilize Windows cash it so even as if the SMB share would fail for whatever reason as long as the images cash into the ram of the PBS server, 1:16:14 everything will be operational. So we're actually simplifying the environment by reducing the number of copies that we take for PS Vita's 1:16:24 rules to adding resiliency and in the Structure. So how does image management work with a plane in the picture anyone in Appling here in the hands? So 1:16:33 good to give me the hands in here. So image management with a player in the main portal management Appliance the elastic lyrics in the user layers now 1:16:42

Appliance, you can export and import that into a disaster location or to a secondary site the main thing with the L. Now is that once you have your 1:16:51 electric wires out there once your user labor set up once you have your letter images out there if the elm does go down those images of school still 1:17:01 running work is everything there is on the album share as far as the assignments and so forth of Json file. So main thing looking at their is 1:17:08 replicating the user layers and the elastic layers, of course at some point you'll have to import that into a new Appliance won't you bring that 1:17:16

planes back and hopefully at some point Cedric doesn't automation around this is going to be in a manual step. 1:17:23 How much is an image take off a workflow which also hear an example and do things and it's a multi-site scenario that case showed earlier to 1:17:36 replicating that Master images of storage application. This CTX article is good for assigning the elastic play repo and also the usually repo is well 1:17:43 that way in a multi-fight scenario where Shane show earlier when you have your own you structure for site, you can point them to the proper location 1:17:52

that way there a casino things from the local dentists in or not trying to go across data center tiers example publishing flow that we've done before 1:17:58 and it was done in the lab. So basically take your letter to me and she probably composed it you create them after BMT snapshot CT DMV published MCS. 1:18:07 You're connected script kicks off and that triggers your protection domain your veeam replication and you broke your protection domain that is pissed 1:18:14 about the scheduler that is replicated to the Dr. Sighn in my case earlier his example, you can go ahead and restore that VM do a snapshot from it and 1:18:21

do your images on the other side that way you have that single Elmo on one side but have that same afternoon is a cross-bite. With the same. Osid that 1:18:30 wait for the user layers. It's not going to cause a problem and that last part can be scripted as well as 1:18:37 we have an agent install have a Powershell script to 1:18:41 create the protection domain. Once that picks it up that script run once you a publishing workflow. And then from there we get that protection domain 1:18:51

and that name is replicated across two are the Datacenter again, you can restore it to the MCS process and go from there. I'll go for this one pretty 1:18:58 quick. The main thing here is that share file doesn't care as long as a data. Is there anyway 1:19:07 you can active passive gslb that have redundancy inside the side and have some kind of files are stored 1:19:16 replicate that as well. I have had customers use cloud storage like a panzura also to the store natively and observe files. 1:19:25

To have that date of Highly available. But the main thing there and making sure the data is there and besides you could do an act active one thing. 1:19:36 I'm testing. I've been playing with your software a lot. So I will be testing this active-active appear software that does that to our location to see 1:19:42 if we can do a to act an active with the sheriff all data on print file Services unified endpoint management formerly known as in Mobile. The main 1:19:49 thing here is if you're not using the cloud unless you are comparing a compelling reason you should be using the cloud hosting service of this take 1:19:59

away the maintenance for that. Take away. The upgrades take away the sequel pieces and then just take care of the things on fire. You know how to do a 1:20:05 cloud connector you gslb for the Gateway Inn and make sure your infrastructure components that are on site like your exchange web apps and so forth. 1:20:13 Are we done it as well? Fun stuff we brought this town original we 1:20:20 had so we broke it off until I say different sections. We literally just had large accountant. So I stayed late endowment just some of the things 1:20:30

that you can you need to consider when you look in a fight a giant as Eliza the green Biloxi 1:20:40 Royal 1:20:50 in the DMZ 1:21:02 lose 1:21:03 his house or this one can reply. Active passive or active actual gsoe proximity gateways for my access to a science. You need to consider that. 1:21:13 I typed in region and then across multiple sites. You need to make sure you worry about that and my soul that I am resilient moving any so I didn't 1:21:25 see a secure viable pairs of gnats guide has he 1:21:34 needs to Spanish Origins sequel 1:21:40

doesn't have to be back to be who's on Travis group. You need that platforms because you need to make sure that 1:21:49 Before you start dropping things out in time to identify exactly the same as external you need presents in both a sentence so that if you have a 1:22:01 fighter of a Dyson set the other day Centrum still pick up down here. Make sure you go molding. When do you send the presents for a 1:22:10 ticket thing you do if there's a great article if you Google authentication using Group Policy to make that 1:22:20

and make sure that you're broken from the right eye center 1:22:29 to see if 1:22:33 you need active classified as what we did is we had active passive dying either way. 1:22:42 Gslb you can about tonight's activates about to pass his time with the same as licensing. This is essential parmesan. So one of the 1:22:52 things we did his refill a controller. in liphook and one in a juul puts it into a Parmesan which 1:23:02 was the the vape shops in desktop parmesan didn't Didn't probation anything from that time but that meant that if we lost 1:23:12

the sense why you still have manageability the phone because of using the right from the other side. This is a recording structure underneath you need 1:23:22 to make sure that the cops think when you change to the ghost, what is 1:23:31 nothing really you have to manage from other than the 1:23:37 resource? Like how do I need to worry about if you're using everything out? Next car that you not going to steal from the 1:23:46 olden everything in the cloud to Great stays up for this is what you use the 1:23:56

video. You can go to write 12 and having to stream died at 3 to clamp connector. The things you need to worry about your image provisioning how you 1:24:06 manage him and your core infrastructure. Right on. 1:24:14 Get seven seconds 1:24:28 to 1:24:29 see the one the one on 2168 100 network is in lip hook. Us if we refresh that we using storefront. 1:24:41 That's running out of my freaking mind. Are we going to go in one of the desktops? 1:24:51 Monroe said can I get a fire out an application? To make sure we have the rock wallpaper on the desktop as well as 1:25:02

important and the SYM number is right. We have 1:25:12 an application and we have a desktop running. I miss Christina Studio. 1:25:22 so if we look up what time she going on behind the scenes for this we had you can see that the desktop in itself is running from 1:25:32 IP vdio to which is actually running in liphook. So Enlighten the date Central and if you look at the application is Ryan from 1:25:42 ipd ir-1 last because we had application to run from the usual Diet Center as a primary and Fire live. 1:25:51

It's a leopard and we pinned the vdi to run from nipple his appointment and file up to his you it has a 1:26:01 coin that Then it will be 1:26:08 able just carry on lotion for the use it so now what what days will start and we can see me like an entire date sent to fight in till Switched Off. 1:26:18 How can I choose a seat beside of the licensing server everything in different? We just shut it down. 1:26:28 Free controllers it down. So that's the one in the primaries empty parmesan that's going on the two Secret Service gun. 1:26:37

and videos I know I didn't hang out in 1:26:47 the open sea. Because until they learn how to 1:26:57 move that we found out. I have a will then then that's just the way it is, but everything else you can. You cannot move around behind the scenes. So 1:27:06 what we doing here is WE Logan on to the next guy in the shop in the Azure tenant. All the Vets wish I was out there is still on you can see if you 1:27:15 guys are gslb order the stuff on Primacy significant is now gone the gslb map states is between the two sides has gone. So that means we've lost 1:27:23

by Dennis on premises, you know, your sister still running. The reason is repentant this too is you we didn't get power of Everything Is You by the 1:27:33 notepad app isn't affected. So the user So if we look at 1:27:41 SQL, this is kind of sim. That's what Shane was showing on it. You can see all those those full bags for the 3 for those who choose not to have some 1:27:51 test results in the table moved up to move out to as you If we do nslookup on the web for 1:28:00 UK Shelf. You can see that's going out to his you. 1:28:10

HALO jump into Syria 1:28:21 tamales the phone. So we still have manageability the phone that's cuz of what how and then sit with this empty parmesan to XD 01 is going XT 02 is 1:28:31 in the same as iron but it's running from as you rather than on premises prepaying work by so you can see the worst vices filed and that's going out 1:28:41 to the clown now. So that's now running from you is your tenant on premises. if we request 1:28:49 log back in 50 seconds in the back and getting nervous. 1:28:59

Let me know what you should be able to see is the the application 1:29:07 subscription database you notice of the Bolton. Oh, so you're using storefront in as you'll need one annual application your favorite. So it's a 1:29:17 subscription stool. But because the 7th group is in the same. It all the shops are in the same group application subscription has been pushed out 1:29:24 to the Clown. So from a user perspective. Ball being kicked out of the vdi which there's not much you can do about is pretty good. 1:29:34

Everything just moved out that you can you can test these Uber play there that you need. We not telling nine seconds and what you can say. 1:29:43 What's the address for questions that desktop application from the gym? That man done 1:29:55 and that is also sorry we don't have time for any questions, but we will we will hang around. 1:30:05

Cackle comments for the website

Buy this talk

Access to the talk “Citrix Synergy TV - SYN217 - Multi-site recommended practices, reloaded”
Available
In cart
Free
Free
Free
Free
Free
Free

Video

Get access to all videos “Citrix Synergy Atlanta 2019”
Available
In cart
Free
Free
Free
Free
Free
Free
Ticket

Interested in topic “IT”?

You might be interested in videos from this event

September 28 2018
Moscow
16
122
app store, apps, development, google play, mobile, soft

Buy this video

Video

Access to the talk “Citrix Synergy TV - SYN217 - Multi-site recommended practices, reloaded”
Available
In cart
Free
Free
Free
Free
Free
Free

Conference Cast

With ConferenceCast.tv, you get access to our library of the world's best conference talks.

Conference Cast
505 conferences
19653 speakers
7164 hours of content