Duration 43:37
16+
Play
Video

Citrix Synergy TV - SYN216 - Citrix Virtual Apps and Desktops tech update (May 2019 edition)

Mayank Singh
Technical Marketing Architect at Citrix
  • Video
  • Table of contents
  • Video
Citrix Synergy Atlanta 2019
May 22 2019, Atlanta, GA, United States
Citrix Synergy Atlanta 2019
Video
Citrix Synergy TV - SYN216 - Citrix Virtual Apps and Desktops tech update (May 2019 edition)
Available
In cart
Free
Free
Free
Free
Free
Free
Add to favorites
453
I like 0
I dislike 0
Available
In cart
Free
Free
Free
Free
Free
Free
  • Description
  • Transcript
  • Discussion

About speakers

  • Mayank Singh
    Technical Marketing Architect at Citrix
  • Thomas Berger
    Senior Manager - Technical Marketing & Competitive Intelligence at Citrix
I am an experienced IT Architect and Competitive Intelligence Analyst focusing on digital workspace technologies. I have over 20 years of experience in the IT industry and enterprise organizations from a wide range of industries, including financial, public and manufacturing sectors. As part of these engagements I have been responsible for full lifecycle IT projects, ranging from pre-sales to post-sales, including strategy, design and implementation. I have a strong understanding of business and IT processes as well as project management methodologies.
I am working goal-oriented and result-driven and I’m able to work independently and within a team effectively.

About the talk

Topic: IT

Stay up-to-date with the latest technical advancements in Citrix Virtual Apps and Desktops. Are you incorporating the latest capabilities and recommendations into your implementations? With quarterly releases, it is extremely important to understand what new options are available, what optimizations are recommended, and how to properly design around the latest technologies. Thanks to the speed of innovation, so much has changed in a year; this tech update session will provide you with the latest information on numerous topics in a fast-paced format. You will also learn about details on the newest releases, the latest results from testing and optimizations, and updated design best practices.Note: This session will be live-streamed during the event and available for on-demand viewing post-event on Citrix Synergy TV.

Share

Write hello everyone. Welcome to session to talk about this with 00:06 me on stage on his first civilization ever. 00:11 Expecting Dan to be here he is mine. And yes, he's so he's got a 10 section of 00:28 the workplace that he's running late and I think we are so thank you for choosing to listen to us. 00:37 These are the hashtags 00:45 make sure we can actually find it and we would love to take all the fame all the kudos for this presentation. But in reality most of the 00:54 things you see is not all work. I mean we put in some work into this lights, but actually it's a combination of of the input of a lot of people you 01:04

see a lot of citrus and please there then fell asleep in one of them who started all of this stuff to external Papa's which contributed in Disguise 01:11 unlock Indy s i n i c t r o f u slides from them and you will see the logos that I will make sure everyone gets to Kudos folder for the work they 01:19 spend into this. All right. So let's get started because we have in good old tradition 2 million slides. We have to go through stop Windows 10 01:27 foundation for pretty much. All I want is a most laboratory TV off vdi deployments in the seabed world and Windows 10 is grade braking 01:36

system. Everything is great, but it has cost of Italy a few headaches and not so much around not so much Roundup application compatibility or new UI 01:45 or anything like that the key problem for most of the customers, especially for customers in a large customers customers and regulated Industries is 01:54 the first release cycle right adopting their it operational processes to Windows 10 has been a challenge picture of him for my own interests who has 02:02 had challenges adopting the it processors for Windows 10 if you who still has challenges Still a few. Okay. Good luck with that. 02:11

Operation challenges of a 3-1 update when is the key things but 02:23 if we take a look at the 02:27 sky Bridge across any going to see a lot of those grass to just walk you quickly through then let me see it the graph nothing new and it is Canada top 02:37 you usually see what we are looking at in this case is going to be Windows 10 and we looking at available memory and what we've been you using for. 02:45 All the tests is Logan vs is the test metric in the back with your tea at using the login at the knowledge worker login vsi knowledge work of work a 02:53

little more performance in some of the workers being 03:02 tested and it's being pointed out. All right, so let's focus on Windows 10 into challenges. This is a Windows 10 1511 release a few years ago. 03:12 And then from there the resource consumption only increased right? So this is the next release next release. Can I hit it? And then we go down the 03:22 only released which kind of went up a bit is the 1809 which is kind of the Lost release. I mean this morning or last yesterday to release 1903 and we 03:30

haven't got around to test it yet. So we will see if that will increase any of that in trees of performance. This is quite a bit. I mean if you look 03:38 at the scale, this is quite a few hundreds of megabytes of change and that on a large-scale deployment has quite a minute. And in one of the reasons I 03:47 was here for this is that has a lot more going on under the hood of Windows 10 Services. All the windows Reese's top to bottom and there's not too 03:57 much and change your default apps with a lot more services in the local scheduled tasks to run how he does things we can do we can optimize and we've 04:04

got to get to that dates on butt out of the box that lot more going on and there's a lot more resource consumption will see how 1903 will perform if 04:11 they change the way you see the gray line. That is the 04:19 number of users looking on into the system and has more users logged in with you to see if you are increasing. These are the versions are going in 04:29 chronicle chronological order and as we get closer to today to get that one fixed issue is Coming with the Lost 04:37

release that's one driver seat at like quite a bit of the different things. If we remove OneDrive that we got a lot closer one drive is really at The 04:47 Forum in talk and it's great if that's a lot of great things. I mean, I think Chef Idol to fix by the look cooler, but that's my opinion, but it's a 04:55 performance Hawk and you need to really pay attention to it. If you want to have it running which is default or is he disabled the same thing is with 05:03 Logan X. So if we look at a similar challenge me local news has the most users on the system the higher the pressure on the system to login times goes 05:12

after its that's what is expected in the more we get closer to today the most resource-intensive Alden agency OneDrive has quite a bit of an impact 05:19 from results perspective. So Windows 10 is afraid of us has a few operation operation of perspective with Elsa from resources Prospectors with me 05:28 to pay attention to that. It would have been to the windows to the world. 2019 05:38 memory and disk. I always kind of up and down but it's that certainly is a bit of an performance. In fact everything. We wanted to understand like 05:47

in-depth. What is the tip of all to 220 16th? Because that is one of the area project our Consulting teams are in all the time. And 05:56 that is the last night in the blue line is just default 06:05 optimization default implementation the Orange Bowl. This is where the new version of thin wires ADHD X protocol underneath it using the 10-yr 06:15 perfect. So they did a few optimization stamp the boohbahs what we really need to look at. Let me go to Windows X 16 Windows Server 2016 BC about 06:24

15% roundabout to 15% ballpark reduction in single-serving scalability. The good news is if we look 06:34 into 2016 to 2019 again another flight from Logan vs. I did it a few tests. That one test was with office 2016. The other one was with of the 2019 06:44 the singles over scalability in packets just 4% So this is quite promising, right? We don't have too much customers get the jump already in 2019. So 06:53 we'll see if that holds true in the real world. What is this a good indication of Microsoft took into account quite a bit of 07:03

difference between Windows 10 and 2016 World take a look at what is the 07:12 actual difference between the server operating system and windows 10 desktop operating system and if you heard about it, but actually going into the 07:22 Windows Virtual desktops option where it is actually a multi section operating system. So if you guys want to know more about which has the option 07:31 that you should choose for your video. Do you have a session tomorrow at 1:30 that I am doing as part of The Geek Squad at the work place where we go 07:39

over all the different options that are available. And if you guys want you could come and take a look at that section be right over the next volume 07:47 so quickly going over the different CPU usage of a task Walker versus a knowledge worker and a different 07:54 optimization start with what's with Windows 10 vs. Windows Server 2016. You can see that the CPU usage is almost half 08:04 every time vs. 2016 Windows 10 is always more but under up bandwidth consumption side. It is almost the same. 08:13

So if you're looking for any work clothes that are more CPU intensive and you want to maybe move to a 12-7 2016 vs Windows 10, 08:23 and obviously since David is not out yet, but we will rock you. Unfortunately don't have a 08:33 key for Windows 10 multi user with the great 20/20 to 30% right impact on over scalability. So that's probably another way of a 08:42 creating a test of delivering a desktop to the use of its everyday usability off application tomorrow and performance can 08:52

also have an impact. By the way that what is it the 10th Guide to the citric workspace geek's Guide to the workspace next door. It's really cool to 09:02 talk to all the different layers and all the defensive line positions. And actually I believe there really people sitting there like two and a half of 09:11 1 and 1/2 days is all the sessions one of the other end and get the full the full thing. It's I think it's pretty cool stuff. Anyway, let's get going 09:17 and let's look into optimization. I mean so far, we've looked really busy graphs and the default configurations except for removing one drive in some 09:26

cases optimization to have to disappoint you again. In about individual Echo actual optimizations. There's a great 09:34 session for that. So we removed it from this one. This is a running later today, Martinsville. VA can rupture lost key of running 45 minutes walking 09:44 through all the different optimizations. You can apply and what the individual impact is on each of them. So if you if you if scalability optimization 09:52 is your thing, make sure you go. I think it's repeated tomorrow as well. I'm not exactly sure but I think tomorrow afternoon we going to have to test 09:59

again. But one thing I was given a few things we have to share any way optimization stair is optimization small fan off in the signs in many cases and 10:07 they think it's just a few years ago. About hundreds of pages of recommendation for optimizations. And everyone has their own little person The Twist 10:16 to it and you can still do that. But it it's hardly repeatable and stuff. So all the vendors golden bed bugs. Should I think that's the vendors I want 10:25 to say created our own optimization engine and Citrix has the optimization engineer behind in case you haven't checked it out. 10:35

This is the thing you should do it comes with a ton of templates and also has a Marketplace and there that you can pick the optimization template for 10:45 your operating system and you'll build you using and it's recommends quite a few things that comes with a few cool functionalities everything you can 10:54 take a template so you can pull up to my so you can take it and paid you can just analyze your system and compared against Template would recommend 11:02 and what you actually have considered you get the kind of a Delta analysis that you can obviously say I want to executed and then you optimize it in 11:09

whatever shape or form you have completed the optimizer for all and that's pretty awesome has a rollback function requires a bit of Polish gel usage 11:16 per day. She can say all the things that just applied. I want to roll back to whatever state I had before in case you doing some sort of an AP testing 11:24 a one-way wasn't good enough then you control it back with that tool as well as the optimizer in general has a very broad Vision. So it's 11:32 supposed to be like a framework for analyzing configuring optimizing pretty much all the components of the Citrix Tech at the moment and it's going to 11:41

be more like a division fluid will be kind of a Swiss army knife everything at the moment like this. It does one thing and it does that things really 11:50 well, which is the really optimizing the performance of the operating system. This is what it's all about. So it's going to grow over time, but it's 11:59 not it's not It's pretty popular already. So it's the second most popular download on support and it's in it's really awesome in case you haven't seen 12:07 this morning Mountain just pushed another version of it's a hole through a few things in case you went to with you always wanted to use your QR QR 12:16

Reader application of your phone. This is the time to do so it gets you straight to the website where I can download a new version of Fortune Yukon 12:26 Denalis on your phone. I'm pretty cool and just to get you an idea of that optimizations actually can do something 12:33 and I performing at what we have here is we test Windows 10 we have for the Baseline we have already remove OneDrive. I'm just to get it somewhere 12:42 close and it can make an impact. So if we run the knowledge book at work load and justify the default settings of Optimizer we get about 50% less CPU 12:52

usage will get about 19% less memory consumption. And 9% less of this lad superformance optimizations 13:01 matter on Windows 10. The same thing is through for Windows service. What we have here is another Logan DSi charger. And what we have here is we 13:10 see the difference. They have also Thunder similar tests like we did. What is the impact of form is in fact, if you go from Windows Server 2012 R2 to 13:20 2016 and they came up with 15% faucet was somewhere in the food chain to 17% So pretty much around the same ballpark if we apply all the optimization 13:28

that we can reduce that impact to about 8% soft imitation a pretty cool thing and they completely can make an impact. So it's worth investing 13:38 time and falls into it. Okay, that's one thing a few things. We have to keep in mind and that is all the tests. We're doing is using Logan 13:47 vs. I look in his eyes really good tool to take out the Baseline and going to test you Hardwell to test on figure out what instance types. You should 13:57 use on a on a on a couch on the couch server, but it but a dream he has a challenge and that it is that it is a repeatable workout. Right all the 14:05

users to use all the same apps in the same way that some randomization in it, but it's not deleting mimicking what uses of doing so all users use 14:15 different types of different type some times that is a pull up. Sometimes that's not an application crashes on loading to make that as well. 14:23 It's not really paying attention to any kind of mystical secret agent spy receive anything like that which might be in a real-world imitation. Usually 14:31 you text Logan vs. I would like a vanilla system you install it at does older all its nachiket require and you don't do the install things on top of 14:41

that. It doesn't pick it doesn't help you understand the impact of real-world situations like a Jetta if a runaway 14:48 process so something runs at 100% all the time for individual use or some sort of excessive looking Stone. This is things we have to wear to keep in 14:58 mind that Logan is Arie just get us just an indication but it never is a replacement for an actual real real test. Customers, which kind of try to 15:07 do kind of a multi-stage testing if you like soda STAAR testing optimizations on you versions of Windows 10 or whatever you want to test like in a 15:16

test bed with this to get an indication of which direction is going and then they dedicate a few users in in production which one that you built. So 15:25 it's kind of a canary roller to feel like it's just like 1% of the US population depending on how pick you up and then they tested and you see how 15:33 much to deviation is from the Chem test with Logan PSI to the actual results in reality and then they roll it out of roll it back and chest dedicating 15:41 a Users on this obviously limit the impact front so you not want to do this and say okay. I tested everything knowing when I run the last thing I 15:49

want to eat when I mentioned here while it's not enough to my station to the traditional sense, but it has it has a disappearing memory optimization 15:59 functionality pills, and I hope you have discovered that already so what it can do to help me that my processes which are running use the seat you a 16:08 more efficiently also that processes which I'm not active at the moment, Florida to disc and don't sit in the act of memory and what we see on the 16:18 screen is a screenshot from an European customer. And what would this help us? There is did the purple line. Is there a process acculynx 16:27

and will be unable DCP optimization for charity to process acculynx goes down. That means we can also increase the actual securitization 16:37 it from the use of perspective what is assaulted in 10 minutes that's a response time with a discussion and the customers actually able to add more 16:47 users on stubble UVM really can make an impact. It's really depends on what you do. I'm in RI depends on how the applications are working really hard 16:54 to get a ballpark number of what you what you can expect you to lie. If if we don't do anything I would say somewhere between 10 and 20% more users 17:03

should be there. I hope you have a French customer. They have some weird homegrown application to get 80% more uses on it. But I'm going to say the 17:11 application is probably not. What are you up to but it certainly makes things better one 17:18 interesting combination because we talked a lot about login vsi. If you'd like and be as I hear a test run into enable CPU optimization of WN you're 17:26 not going to see an impact and the reason is the optimization on the top of um system and it's happening on the schedule. I think it's every every two 17:35

minutes of every 5 minutes or something that it does something but looking PSI is opening and closing applications at a faster Pace solely me to text 17:43 something when I talk to myself that state has already changed. Obviously not how users work to keep tend to have an outdoor open longer than a 17:51 minute, but that's so you're not going to see you looking to the side end. It's been contesting this. All right, let's go 17:58 take a look into what virtual desktop releases. What the hell is he? In fact is that what we have here is 18:06

again Windows 10. It is all the different versions of apps in desktop until 1903. And 18:16 is it a quick and we see one couth Trent it's going down. So I'm going down to reason as long as they were going to see something and late around 18:26 that later on if we did a lot of investment in the HDX protocol more efficiently going to see if you examples of what that means and what we did under 18:36 the hood to get them but it's also we take a look. We took a long hard look into all the other services to VTA brings and distance from the CPU 18:44

perspective. It goes down for the talk of work load the knowledge book of occlude same thing because it was coming down quite a bit. So adopting the 18:52 latest current Lease Only Ohio Skate ability on the system and if you look at memory the same thing 19:01 Windows 10 same workload and we can see a similar thing. Although not as pronounced as with CPU vs CPU. What's the what's the coal thing we did 19:11 in general or about 25% Improvement when moving from the 19:19 715-022-1903 current release. So this is pretty please pretty cool. But as nice as this is us like we got from ITT are down there. They did a similar 19:27

test and it came to a similar conclusion which is the left is 7715. It's the blue in the orange and on the very right inside the doctor with this is 19:36 1811. So it between those versions at about 30% more use a kitten to assistant. So it tastes really scalability is an issue as 19:46 a topic for us and we re in Destiny 2 when you adopt the latest, you might get some of the Home inspector you lost when patching for Spectre meltdown 19:56 patches just by going for the latest you alright sizing moving into sizing really quick sizing again is a notch and 20:06

I said that they always one of the most asked question is how many users do I get to the system closely followed by how many men how many penguins 20:15 does my user requires? Why does to pass it always come up and that's all sorts of magic. You can do one thing leads when Talent came up with is 20:23 the rule of five in ten. This is pretty simple, but it's amazingly accurate. So I said she heard it from rule of five in ten works is if you want to 20:33 figure out how many users do you get on Virtual desktop Austin up system. Can you just take the calls you have in x 10? So we have 28 cause you 20:42

multiply 10 by 10 you will end up somewhere in the 280 users per machine if you do it for you 20:51 just * 5. I'm in the slides from Nick and Jess actually run the section recently, but I walked through all the different are 10 examples of 21:01 real-world customer and he was I think the actual average across all the customers was 5 and 9.7 or something to put the world record 21:10 the digit 5 in 10 days really amazingly accurate. Now that's one very important thing when talking about sizing systems in this is how to see if you 21:20

that we won't slit recently. So in the past but not that long ago since he was where I had said anything which was called new my nose. If you if you 21:30 look at it see if you you have to call each call has a memory and does a bus in between as we add more calls those calls all use the same bus to get 21:39 to the memory as you go beyond a certain point that boss becomes a bottleneck. What did the City of Angels do they split the Sea View Inn half do you 21:46 get 1 Newman road which has a witch multiple cause she a boss and a piece of memory on the other side of the same thing. Now if you happen to have 21:56

like a VM, which has more CPU requirements than one of those new my notes can do then. VN spilled across to the other Newman Road. And you said you 22:03 had a very slow memory access photos Kohl's Sonoma thrashing happened and ended performance went down quite a bit. So you really you had more 22:12 resources performance goes up at some point. It actually went down that is because that memory Texas was too slow for Skylake processors that changed 22:20 their now have a mesh and that measure allows fast access for all the calls to all areas of memory. So this restriction Pend Oreille. What is a 22:30

recommendation for sizing if we start with something this is what we do for me the I am not going to read it out for you. I think this is just a 22:38 starting point for Windows server is what the important pieces for CPUs. We always size for one full Numa Numa node not going beyond that and not 22:46 using sum of numbers because otherwise you might have this new my thrashing happening. This is for the old CPUs, obviously for newest Infuse this 22:56 whole thing goes away. I mean memory sizing and stuff is still there. But the CPU piece it's going away at them now resides around operational things 23:04

are we think we can create a massive message? We almost have like hundreds of used on them the challenges what happens if you going to maintain it or 23:13 if it goes down for some other reason to tell your domain is huge. Do you want to keep it small so then it really comes down to you or operational 23:20 requirement what you would like to do? So it really is your thing. I think at the moment. We don't really have a best practice but we but it's a 23:28 app. I think we just use to starting point spread in the car. So we have a big question is on-premise is nice. But how do I 23:37

size look how to say Novia celosa across all the different instant title and the F-16 serious holds the most used for what is 23:47 what is the cost associated with each user? So if we take that and X the cost and we end up with this and at the end of the day the f-16s is our 23:57 favorite. The reason for that is to get the most bang for the buck and each user is somewhere around 2 to send flower in Disguise toys pretty pretty 24:07 awesome. So, how about HDX? You got a few more about HDX. All right, so 24:15

I'm here to talk to you about all the optimization that we've done with HDX and how it actually impacts the use of performance and how you can make 24:25 your users get the best students. Even if you just three policies inside of the Opera in Sarasota qualities. So the first thing to note is that 24:33 applications themselves are not constant amount of traffic that can cats get sent over the wire or even a single application could vary based on what 24:43 you're doing with it. So, for example, when you presenting like we are you can see the all the spikes if you're creating a PowerPoint. It's flat 24:53

amount of data that's going through is very less than any any idea why this biking anybody nope, 25:01 it's just moving one side to the other. This is the amount of data that's getting pushed across the network when you just move them from one side to 25:11 the other. So every time that you see something that's happening on the screen. We have to change we have to compress we have to send it across to the 25:19 network and that is what makes it different looking at average bandwidth usage on 2016. Every application is different. So if you have 25:26

PowerPoint with word, you will see that across-the-board on every version. What is using much much lesser? Why is that because we are able to compress 25:36 the content which is there in word cuz of a different technology is much better than the images that are there inside of the PowerPoint presentation. 25:44 So what we would like to give you to take away is that this is a set of policies. If you have the two different use cases one that you want to give 25:55 your users the best performance that does the user experience set of policies and one where you have a low latency. That is the one I wanted the 26:04

qualities. So we try to reduce the amount of network when we have a policy and we try to put as much data on the best Clarity across when we want the 26:12 user experience policy impact of these policies and limitations are so Average bandwidth per 26:22 app again. What? Can you get? What can we do on your left and policies on the right? As you can see 26:31 we've got 35% reduction in the amount of data that is being sent in a gun policy. Coming to the 26:40 next phase of it. We don't know if it's a set of protocols. All of these protocols have individually to find each 26:50

of these are built my separate way today. I'll be talking to you more about the adaptive display why that makes a difference. So in there now, 27:00 we had this one thing called 10, yr. And that was the way that we used to do the video compression as well as image compression on the screen, but now 27:10 he moved away from there or be using that as one piece of it as well as putting in what we call is adaptive display. So what is adaptive display 27:18 is our way of understanding what is being presented to do the user and what they sing in the screen? So this is a web page standard webpage. 27:26

So we'll look at what all it is made of The first thing is in the green box boxes here to video. You've got in the red 27:36 jpg images as well as start a text in the blue, right? So if he just turned and we compress everything in the same level you'll get 27:45 very very different results will be much more. So we understand what is being displayed on the screen and be able to 27:55 compress it based on that and we use different protocols for each of these sections to h.264 is our primary method for doing compression for video and 28:05

Graphics that will give you a really good video performance be used in LaPlace, which is the maximum compression for images and then we go with 28:13 overlay of lossless for text to it's not bloody at all. You don't have pretty and you're able to read text on screen. As you can see 28:23 text is lossless. What happens when we actually move to a screen that's completely text. Everything needs to be 28:33 compressed individually and Santa Claus, right? Well that would cause a lot of data to be sent over standard Financial to screen people. We looking at 28:42

lots of lots number. So what we did was something called wrist detection Theory figured out like this piece of information is being repeated across 28:52 the screen and we'll look at that. We can probably understand that this doesn't need to get sent over again and we are able to replicate. On the 29:00 endpoint there for you get 16 x better rate of transmission Soul II vs. 4 seconds for this particular screen and you get 29:07 91% less bandwidth and 30% better than even RDP and its own 29:16

customers who are in one scenarios will be able to have a local experience 29:23 rather than one where everything is dragging or it's moving slowly when you're scrolling up and down a vexatious like this. Another thing that we have 29:33 built is h.264 Bill to lossless. This is a really cool technology where you would you have user that wants to do in graphics-intensive workload, but 29:42 then there's a lot of interaction that's also happening. So if he tried to present everything in the same quality while the user is interacting with 29:52

the model, then you're pushing so much data that the system itself is not able to get that amount of network bandwidth and push the same amount while 30:01 their users interacting with the model reduce the quality and when they finally actually end up in the point when they want to look at it that is 30:10 where you take it to the complete Pixel Perfect Picture so she can give you is almost to Exmore interactivity or 30:18 the same interactivity the images that are being built will build up to the purpose of a scenario with in half the time. 30:28

And that is available in 7:18 or later. So let's see what the impact of the CV releases are on the actual. It's TX 30:37 performances being sent for the knowledge worker and we're using the u.s. Policy here. So as you can see 30:46 individually each of them are almost the same and we will try to improve them as we go and finally one is the best flight 1903. We've always 30:55 been improving and you'll see why in the bed again for the college World Cup. 31:05 Know when you said the gun policy as you can see almost half of the content is being at the banquet is being consumed because of the policy reducing 31:16

the amount of content that is being sent know. We are adrift cashing in 7:13 and we are discussing the version 2 in 7:18. 31:26 So now let's take a look at it right now. We sort to know everything that you saw was a knowledge worker. Let's see the task about the workload with 31:36 all of the workload in vsi is more towards XL sheets and yard open an email management and as you will see as soon as you put 31:45 cashing in it drops a hell of a lot. Because that's not that all of that input that was going through just 31:55

wiped out from the screen. So the cash conversion to that's where you want to be. 32:04 So I ended up in front sport 32:12 adaptive display works on top of what we call adapter transport. And the ability to use is what I call and light and Air transport which 32:22 is using UDP or reliability PSG college. But what happens in a scenario where the user is not able to get UDP because there is some 32:32 way that there's a firewall rule or where is the portal block and you know, you can get you to be so then if that is the case then the user will not 32:42

be able to get the screen at all. So we looked at Trying to make a connection from the endpoint to the video using 32:49 ATT if that's failed multiple times, then we would try is happy a synchronous process and it would slow 32:59 down your actual logon times as well as make the use a weight for for this to happen. So what we've done in version 2 is he actually added 33:09 that is embedded simultaneously tries to connect via TCP as well as ret if in any case it doesn't find TCP. I'm an idiot will connect 33:19

Uconnect Rarity if in case that you get the option to connect wired ET then seamlessly transition to eat 33:29 and therefore you will get the benefits that we talked about. And you need to know understand that none of this is ever visible to you or the 33:39 end-user. It's all done automatically and adaptive. So it looks at the correct it looks at the conditions. Does it automatically 33:47 next as we see the single server scalability between PCP and nudity? There's no difference 33:56

either one for you. It makes no difference on the banquet side or do you say is that is a 15% 34:04 requirement more on the van side because there will be some requirement for transporting the data based on the it being a UDP 34:13 network-based Network and taking more data. When's it doesn't reach there. All right, last but not least 34:22 something else. What could that be? Everything that is using a standard 34:30 pipe? What's the best conditions possible? Okay, which is the best as in the last 30 years. We've been working with something that bass lines to do 34:40

Mbps as an equity requirement. In reality. It always changes the content that you're pushing somebody watching something else 34:49 on the on the on that particular pipe or something else happening during that Network conditions change and We've added something called adaptive 34:59 truporte sure. We're actually looking at what is current state of the network and changing the protocol as well as a network file compression in 35:08 parameter is to make sure that the user gets the same interactivity and the best user experience. So what really happens in a network? 35:17

We're also watching a YouTube video and trying to do the same work. Your buffers are too low and the amount of data that we want to push through at 35:27 the same time is not going through and you have a bad experience in a scenario where you have a really great network, but we're not pushing enough 35:35 data. You're not quite as well as it's good to reduce the amount of data that's getting pushed and for lower video quality floor or 35:44 interactivity. So what do we do at the current Network conditions? We measure round trip every 4 seconds and 35:53

we have we have a default value of 400 if it goes over 404 for consecutive times, we will drop the output refers to a 36:03 110 based on the other test. If it increases then we'll put them back up and all of this is again underneath the covers. You don't have to worry about 36:12 it. We will do it for you. Make sure that uses So what does this actually translate to in real world? As you can see without 36:21 optimization everything is below the to Mbps smart. None of the data being pushed can actually make it through even if there is spy available in the 36:30

optimization with built-in you could push a lot more data. So as one of our customers have told us they have a 36:40 roving operation where they need to go to 22 different locations across the Earth and all of their data needs to be 36:49 managed at the headquarters and they actively look at what is happening in the real world with 1080p 60fps video con 36:57 without this is not possible at all with this being able to push because they've got a really big pipes. They get 1080p simultaneous 64 37:06

60fps performance across all of their uses. So what we look at with the adaptive display, the other Trooper 37:16 temperaments are if you're trying to copy of 495 from the server to the client to get 2x benefit. complaints about almost or 37:26 617x Improvement All right. Oh, yeah. I'm the Copa Santa from hell. I added a nice surprise to lie to 37:36 you and thank you notes when we said we do 37:46 a lot less always with required Loveless spending to use when going into cloudcompare as compared to all friends, if you would like how 37:56

the hell did you get to that this is really a simple test looking PSI again in this case. I think it was the what was it? I think it's out what you 38:06 see is on the left hand side 54 3v to instant to Bluebonnet Citrix to Green valances VMware and with a blast protocol 38:16 which really forms that it has a great graphics air quality for video content. But as soon as you go off the video content into any kind of text, they 38:25 still use the video Codec h.264 codec to include everything and issue 64 and coding is really really really performance intense. And you've seen my 38:35

tank with all the things how we decompose the desktop. They don't have this intelligence at the moment. So it's all or nothing for them which results 38:44 in a significant role so we can get almost double the number of uses a system in cloud. And hence. We have half the spending the back to you. Thank 38:50 you. We've been talking about a lot of things that have been only to 39:00 do with cvid. So what do we do when we actually 39:10 integrate with our partners product like seven variable to have the ability to give you even better performance. How do I do that? 39:20

Let's take the example of an Enterprise internet. Normally what will happen is everything that has been going to go swing going through the desktop 39:30 has to go to the mpls goes back through back channels to the data center and then hits a site and therefore infection and all 39:37 of those procedures everything needs to go back back harder. Traffic. Is that a single bite that gets flooded when things that 39:47 don't really need to hit. So what we can do if he brought Steven is the ability to understand What's the traffic is so we can 39:57

see exactly into the icap protocol and figure out what data is being sent and how it needs to be optimized as well as we can multi-channel that data 40:07 across the mpls across a mobile 18 or directly over the internet and what these two devices will do between them. They will actually be 40:16 looking on the on these different networks and understand the current Network conditions and be able to figure out which is the best part for the 40:25 stated to be going over, right? So what did happen really is that? Be able to see the icon priority for each of the 40:34

individual. Kind of content that is being sent so audio, which is the most important real-time communication that needs to be the highest priority 40:44 your graphics and video which is the next priority as well as if I transfer and printing in different priorities, so we figure out where this data 40:52 needs to go under what channel and we're able to actually figure out without having to do any changes on the cat set up the ability to push 41:01 data across wherever he needs to go as the best Channel possible. If you've got enough bandwidth will send everything in Paris. Otherwise, we will 41:11

go so that would be great value and that you will get better together lyrics 41:20 Awesome. Let's bring this home. I'm quick summary. And in the interest of time. We keep this really short. So Windows versions upgrading is great, but 41:29 it has a skill ability impact pay attention. If you up to my stone table of two minds always keep in mind. What about to my station you had removed it 41:39 gets away from the standard and only the standard is being tested right Microsoft is not testing any any crazy of limitations. Neither. Do we do the 41:46

more you talk to Mike the mole the chances you get into consideration area, which is out of bounds. And if there's a problem you have to revert back 41:54 to the sender to keep that in mind. It's always like this trade of like most of us vs. More challenges with troubleshooting. 42:02 If you don't if you haven't tried it too far and we 42:08 always at in the engineering department work constantly on making HDX better. They have a lot of that's a lot of amazing wizard them in the 42:18

Great, but you have to adopt him that they just current release write all of that. It says it's going it's going there an active and the combination 42:29 of HDX next event in our ability to look into the network characteristics as well as the HD XR. Traffic is is a is there any combination with you 42:36 other than the really can can provide? All right, if you haven't seen text down yet. This is what all the cool technology at all day all the 42:44 documentation all the guidance around design and sizing and what nut goes there to rule of five in ten point Temple. All of all of that is a document 42:53

that there and it's actually very soon is being open to public to contribute. So keep an eye open until 10, if you like the graphic of The Undertaker 43:01 go to the Citrix puth and you will find a few of those stickers there and he speaks guide and session about multiple ways to deliver 43:08 desktop is coming up tomorrow Delmont the old next door to go there check it out and teach rate to session. Thank you very much. 43:18

Cackle comments for the website

Buy this talk

Access to the talk “Citrix Synergy TV - SYN216 - Citrix Virtual Apps and Desktops tech update (May 2019 edition)”
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 - SYN216 - Citrix Virtual Apps and Desktops tech update (May 2019 edition)”
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