Duration 37:19
16+
Play
Video

Citrix Synergy TV - SYN202 - Under the hood with Citrix Analytics

Mathew Varghese
Director of Product Management at Citrix
+ 1 speaker
  • Video
  • Table of contents
  • Video
Citrix Synergy Atlanta 2019
May 23, 2019, Atlanta, GA, United States
Citrix Synergy Atlanta 2019
Video
Citrix Synergy TV - SYN202 - Under the hood with Citrix Analytics
Available
In cart
Free
Free
Free
Free
Free
Free
Add to favorites
283
I like 0
I dislike 0
Available
In cart
Free
Free
Free
Free
Free
Free
  • Description
  • Transcript
  • Discussion

About speakers

Mathew Varghese
Director of Product Management at Citrix
Jim Regetz
Director of Data Science at Citrix

About the talk

Let's open things up and see what makes Citrix Analytics run. We'll explore core platform capabilities including self-service analytics, search, reporting, real-time analytics, and custom indicators. We'll discuss our approach to machine learning models, feedback loops, and deep learning. We'll also explore data security and privacy (GDPR).Note: This session will be live-streamed during the event and available for on-demand viewing post-event on Citrix Synergy TV.

Share

Place and everybody and welcome to Citrix analytics under the hood. My name is Matthew. I am a platform product manager and I'm joined by Jim was a data scientist at 1. So we're going to go over the mechanics of how analytics work. We'll talk a little bit about you. No data protection data governance, but mostly about machine learning and how we using that to deliver inside to you. I know that I'm not the last guy standing between you and the beverages at the party. So I'm not feeling too guilty about it this time last

time I was so I hope you enjoyed your presentation quick question. So how many of you are up security admins are in The Walking The Walking secops infosec Creek showtimes. Pick up one person couple of people awesome and how many of you are on work space today Citrix workspace. Fear of you, and I hope that by the end of this presentation a lot more of you decide to start using workspace a quick saying if you're if you're treating which I wish you are ya use hashtags, and I'll get right then so Citrus analytics for those of you who've

probably heard about it. Tried it a few times. This is Bentley how it works. So all citrus products of an instrument to send events to Citrix Cloud. So any Citrus product, which is a cloud service some of the on Friday on Prime products as well. If you have hybrid right to know how that works. They all send events back to Citrix Club. Now those events are in a process using r a i n m l l got to provide insights back to you. So if you look at the top of the diagram Guard Security performance and productivity, so depending on

which service you are using you get different kinds of inside. So if your insecurities analytics to get security in size during performance, you get performance inside Yoda in operation to see a lot of the operational data product and the control plane stolen data Control Data reapply Ai and ml to it. We look for patterns to look for anomalies and we provide Insight depending on which service you are on a little deeper so far today, we're going to dive a little deeper and show you how these things all come together. So let's start with the bottom

three key Concepts and every level. So we're going to cover three levels today and at every level we're going to cover three key Concepts to the first level is it is there is a platform itself. So there are three key aspects to it. We have eternity data sources, which is the key Concepts. I'm going to find something to talk about then we look at the event of the event format and then last but not the least we're going to talk about Vantage points and data sources. Let's look at how analytics works at a very high level of analytics has some key Services right to the

top right but to get started with your data sources. So when you go to settings data sources and you turn on data sources and makes an API call to the cloud control play all the analytics compatible products and services that you on and then a product list is generated and displayed back in the UI. So the reason why the operation is because you don't have to worry about figuring out what you're in a horse names. Are you excited IDs all of that stuff you just click a button we are to discover all the

services that can send data to us. You don't have to go figure out which Services can send data to us. We take care of that for you and instead we display all of them to you at sidecars. To the person customers last message, you know, if you do analytics you seem to be holding up all my data you seem to be analyzing it what's going on. The first thing I want to show you is that we do not touch your data without your consent. So when you go to data sources, we discovered all your data sources for you. Data sources might be sending data to Citrix Cloud because that is essential to the

functioning of the services but analytics is not turned on by default for you right to what you need to do is go there. So no data processing when you turn on data processing ready to receive data that's coming from your mom and say it's not a pull mechanism. The Citrus products had to push data into Citrix analytics data processing data processing turns on and when your end user start doing something beta starts flowing in barcharts light up, but the key thing is you can turn off data processing at any time. Right. So

once again, this is a donkey service. You can go turn it on at any time. And when you turn on data processing and Sony the end that you start processing of data, and the reason why is because you don't have to go around insulting agents and sensors throughout your environment. We we we just make API calls and we start receiving data because you're so that is key concept. Number one II concept I want to talk about this event was concerned. They have is

God's plan. It's super expensive. I'm sending them all my logs and faces. What is analytics do isn't that going to be expensive for me as well? So unlike other general-purpose analytics platforms. We have custom instrumented every one of her subjects products to send this event. So we are not pulling logs and traces trash dump score times today is an event that looks like that it's a song and the event gets sent to us when an end-user. Something. So there's no stream of events at any given point in time. We only get an event when an end-user. Something

looks like that included in Jason and these events are extremely small and small and because it's their structure structure is well understood by the machine learning algorithms that Jim is going to talk about the system is also extremely fast and accurate. So that is a second that was a second concept. So basically we are not we are not hoovering up data from your you know, from your environment event the events get sent to us when the end-user does something and the event comes to us as a Json object as a Json payload, not every event is about 223 KB

to do if you have a few thousand employees doing something you're probably going to be getting thousands maybe millions of events every day. But because event is just about two or three KB the total data. Ingest is pretty low. The last thing I want to talk about before I hand it off to Jim is vantage point so that way to think about Citrix products is that they are deployed in KeyBank is pain in your cervix environment. So you have the wall space Apple receivers running on people's Sandpoint devices. You got you know, what the networking

products are deployed in the network. You've got two stops that are you know, accessing your apps and desktops. You got content collaboration that's looking at all your files to basically Citrus products are deployed in key Vantage points in your environment and all of them have been instrumented to send events back to Citrix. So basically run an end-user for example shares the file we get an event from content collaboration with used to be called Sheriff. When when an end-user Android device we get an event from uem what used to be called T-Mobile. So basically you get

this 360-degree coverage when you use Citrix analytics because all his subjects product is an event to do citrics analytics. We use these events. We consolidate these events to create user profiles and that's what Jim is going to talk about to just to conclude that section. Basically at the bottom of God don't get data sources events to analytics and these events are coming from Key Vantage points in your environment without them going to hand it off to Jim was going to talk about the use of profiling and most importantly machine learning. So now that we've got these events coming in

from the onboard of sources. We actually want to turn that raw data and something that's actually useful and I turned that into some kind of intelligence and information that you can use at the business end of the product Social Security would have a couple of main things here. We have user profiles. We have whiskers and we have risk indicators. It's going to go through each of these now the other thing in this middle layer where the magic happens. That's also where MLS and it's not exclusively a mouth. But that is what we have in there. And I know it's kind of a slippery context

concept, you know what to do. If if you're not a ml practitioner probably have an idea what it is, but you've heard lots of different ways of describing it. So I want to be clear about how I'm going to talk about it and hopefully I'll convince you that it's a good way to think about it in general and you'll bring that forward when you hear about read about ml in the future not just today. So to do that I want to go back to to what's in my mind. Probably. The best thing we have is a kind of textbook desk definition of machine learning in this comes from some academic Publications from

a decade or so ago and you actually see a variant of this in the Wikipedia page for machine learning if that's any validation and it goes like this machine is set to learn with respect to a particular task if it improves performance at this task following experience. Okay machine is said to learn respect some job and get better at that job with experience. So if you take some time to internalize this I think you'll find it so useful statement, but I don't think that's easy to do and it comes down to some of the language in here. So I want to do a couple of things before we move on. First I

want to take the word machine, which I'm telling you. I've been doing this for years and I still think there's a part of my brain that conjures up an image like this when I hear machine and machine learning could just happens in my head and it's distracting. So if this happens to you or any case when you hear machine learning mentally cross it off and replace it with software because certainly for any context that's relevant to folks here in the room or watching online software is what we're talking about here. Secondly the experience part personally. I find it hard to think about software

or machines experiencing things. Right? So if I go for a run in the beach wearing a smartwatch, I'm definitely experiencing that right. I can feel the sand and smell the ocean Aaron feel that burning my lungs but is my watch Dancing that you know, what's happening as it's got sensors on it that sensor two sensors are detecting things about the environment of my physiology and sending that is data much like we talked about Citrix analytics time processing center. So I think let's forget about the following experience thing and just say using relevant data because that's what's happening.

And then lastly will take the learn verb itself again, you know, this is something we do is people but it's hard to think about software learning unless you've already become comfortable with the concept. So is drip drop that part and just say apply. Ml which leaves us with the statement that software is said to apply ml to a particular task if it improves its performance of that task using relevant data, and that's really hit. So hopefully at this point you're thinking of us pretty simple, but also that is pretty powerful and everything I've done here is take ml away from a particular

set of techniques and tools and turn it into something. That's the software capability. So ml is what you have when you have software that's capable of getting better at doing what it's supposed to do with data. To do that though. I'd actually Implement something like that. You do need various techniques, you need algorithms any of those methods and and this this diagram here is not an exhaustive said it's not that you know, comprehensive taxonomy of machine learning but these are all things that in our team we've researched right experimented in some cases implemented and

again, not exhaustively but things like various unsupervised anomaly detection methods which is important for security deep learning been very hot for a couple of years can do powerful things. We've explored some things there and then even statistical methods which I put on here and I might seem kind of quaint talk about statistics in the modern machine learning era, but the fact is if you're using the Machinery of Statistics to better understand some phenomenon in your data that allows you say to make better predictions in those predict predictions lead you to some more performance out, that's

measurably better with more data. Then you're doing machine learning, right? It doesn't matter that. It's not Deep learning is still machine learning. So the last thing before we get back to the product, I want to introduce the concept of the data pipeline. You may be familiar with this but the basic idea is what we've taken math diagram and just flipped it on its side. We've got Robbins coming in and going through some Transformations and ending up on the stuff that you see in the product when exposed to apis and does cool stuff for you. An important part of this is that unlike an oil

pipeline. That's just moving something around in an ml in a data pipeline were applying Transformations along the way to turn this information into something more useful for some reason. So all of what we have in Citrix analytics is built on data pipelines. I mean the stuff is implemented in Azure and we're using various Big Data Technologies of flow this through but fundamentally again this idea of having transformations of data along the way they do something useful is what we have underneath the hood. I want to highlight that there are two different kinds of Transformations. There are

what I call the traditional meaning on ML ones or the logic in that Transformer is hand engineer right as written by a data engineer or data scientist or software engineer and what it does is basically the same all the time until we go back and fix it and update it later and we can do some really valuable things with that but it's different from learning capable transformation where we're deploying a model with some kind of trained algorithm which itself is adaptive and responsive to data and maybe adaptive and updating as the date is flowing through or maybe that Atmore, Liam.

There's some other kind of almost a separate pipeline is one pipeline where this transformation being applied and a second one that's got a learner that knows how to take some other data and use it to update what happens inside that thing. So you get this kind of dynamism where that Transformers changing all the time as a consequence of data and if they'd is leading into perform better. So how does that all fit in when we go back to Citrix analytics? So we have these pipelines. In fact, we kind of have many parallel pipelines is not one linear thing. There's front of a whole bunch of

things that do. Different jobs. One of them is producing user profiles. This is no ml in it, but it's still super valuable. So what happens is events come through we detect those we see who does events are coming from as we discover those users. We throw them to a page in this is accessible through the apis. Well that list all those discovered users and provide some basic information about them. Like what types of data sources they came from how many you need to buy so they've had unique locations and so on over some specified time. And then if you click into one of these like Josh Carpenter

here third one down. You can go in a Jostens page go into the user info section and see a little bit deeper information. If you have AT&T gracian to get directory information, which is really nice. You could see who tosses what their role is provide some nice contacts for interpreting what they're doing and a little bit of a deeper dive into the specific applications and devices and location. And the number of access is associated with those but now it's back out. I think more about Security will back out to a different view which is Jostens timeline. So here on this timeline, we have

specific security relevant activities and events. And we call these indicators risk indicators. It's a core part of the Citrix security analytics product and the key things to know about these I'll go into some details again. They happen they occur. We create them when unusual things happen by users and I'll show up in this timeline. Once they show up there you can click into them again. There's no ml here. We're just providing additional details that you can look at. So for example of Josh's had access from the new device you can go in and see what those devices were and when exactly

this happened The last thing on this page is the risk or so that's in the upper left here. Just as a high-risk user score is 98 700 and I don't want to go into a lot of detail about the risk or but think they have to know or this first 50 scores are unique to users in the changes in those scores. Are you need to the users that basically when the user does risking an abnormal things that score is going to go up and it's going to go out kind of as a function of how many of those things they've done the intensity of them and end how frequently how close in time so there are

various kinds of ways that we amplify the score based on that and then if user thereafter doesn't do anything unusual for a while the scores and they Decay back to a normal level. So this provides your way to know where to look in first place right now. I'll have to dig around and look at all of your users. You can look through a structure of you at the riskiest users and start there to investigate what they're doing. Okay, so I said to hold off on the risk indicators themselves. So I want to take a few slides to go through these and I've oriented oriented them kind of round these four

categories that you're not going to see this in the products. We don't even really talked about this allowed within the team, but I think it's useful just to structure an understanding of different types of indicators from a simpler to somewhat more complicated and it kind of gravy so start with the simplest ones that direct indicators of compromise and these are immediately observable security relevant events and properties of events in Citrus analytics. We don't have to do too much other than recognize when something happens, like for example, if Citrix Gateway has detected that users

device has something strange going on when it's done an endpoint scan and sent us that failure event. We know right that we don't have to do anything else similar ways to fix Access Control based on URL reputation tells us that a user visited a risky website we get that as an event. Citrix endpoint management to text to the users to buy some become jailbroken again, the logic happens they are but then point management project products and we get that as an event. So in all of these cases were able to pass that through we don't have to do anything else and what's nice is that

sometimes it's just preferable to put that kind of logic which itself me about machine learning on the product itself and then it heard it through Citrix Analytics. Second are the drive behavioral analytics. So let me unpack that when I say analytics. I'm talking about not the broad pursuit of analysis, but plural of an analytic which is some intentionally computed thing to measure some characteristic in our case about user behavior. And I stayed arrived because you can't just look at one event and see it automatically you have to look across multiple off and many of

them to measure this Edition importantly you have to have learned something about the user or the or are all users to contextualize that and make a determination about whether the behavior is risky and that's where our learning comes in. So I'll give you two examples one is unusual login locations. Where we ask a few things and when I say we asked a few things that analytics is built in Texas stuff. First is a simple one or have you been here before? I just a little learning is basically being able to keep track of where you've been when and have a kind of memory components that

if it's been two years that's different from if it was yesterday and it's actually we asked are you typically mobile? So if I always login from my house never anywhere else, but suddenly I Pier from here in Atlanta. Where is Matt bounces around all the time and come see her will regard. My behavior is inherently risky or because it's unusual for me to move around. Additionally. We asked if this is a popular place in your org. So again, if I'm somebody who doesn't move around much and I have an event that comes in from the location of Citrix headquarters turns out because we we look

at the behavior of the locations within the org we can detect that's not an unusual thing for me to do compared to showing up in the city that nobody from Citrix has ever had a login from before. So all of these things is what we factored in to the analytics and used to establish. These pretzels are what kind of falls out of those bounce a flag is what would be considered unusual and therefore potentially risky location for my login. A second example very similar to this is with stitches content collaboration where we want to measure users who have excessive download behaviors again

relative to the Past download Behavior. So we need to ask and track overtime with typical for a user be able to kind of measure the distribution of normal download behavior and have sonography McQuaid identify. If somebody does something excessive compared to that which may indicate a data exfiltration event. And additionally we say, what's normal for the org. Sometimes we don't want to be in a case where if I download a single file and then 6 megabytes in megabytes in a 50 algorithmically by itself. That looks excessive but we can text the last we see people do across the entire or

we see that That's Not Unusual on we use that to adjust the risk or as well. So that's the second Factor. The third one I want to cover. Glycol purpose-built attack protection so so far. I've talked about things that are kind of about a device or user security posture and some behavioral components that may indicate something suspicious going on, but they aren't protecting particular attacks. Now in this case and example we have in the product today is around ransomware detection. We have some variants of this we are specifically trying to detect if ransomware has

occurred and and we do that with Citrus content collaboration if your user has a sinkhole enabled and that sing tool is is basically monitoring the files that end up getting hit by ransomware. What's a the ransomware encrypted copies of those files and delete the old ones because that's happened to the files. The sink tool is going to do the same kind of operations. And because the same tool is one of our onboard a product we're going to get events in Citrix analytics. And so there's both a signature component of that and we've actually learn this by taking by creating sandbox environment

where we've we've actually run ran somewhere and learn what is Signature of the ransomware is in terms of the kind of synchronize patterns of spikes and upload and download behaviors on his files and that in itself is pretty cool. But it turns out if we then apply that's a real data users can do normal things that kind of look the same way. They're almost indistinguishable indistinguishable from that ransomware. So what we do on top of this is learn again was normal for the behave for the normal behavior for the user because with actually really rare is for somebody to get hit by ransomware

who also just happen to be doing things. That kind of looks like ransomware. So by combining these two things we learned was normal for the user in terms of upload and delete Behavior look for excessive behavior in those regards and make sure that it matches that signature of ransomware that we can detect when things like this happen in the daytime showing here was from an actual case where we detected ransomware a user who had a ransomware attack. All right. So the last one I want to talk about I'm really excited about it. And I highlighted is Future and every language at the bottom to

emphasize that I'm making a commitment about when you'll actually see this in the product and left arm, but I will say this is something that's it's not just an idea. We actually are running this kind of in the dark underneath and using it internally ourselves and I'll call this kind of a more open-ended anomaly detection. So again, the previous three were all about something that directionally or just a priori we know his security relevant here almost as looking to see what falls through the cracks terms of anomalous Behavior. And I can give you some intuition for how it works. Take

these three pictures here with your all pictures of a person's face two of which are real photos of real people one of which is a fake. It's just a painting. I'm sure everyone of you can easily see that recognize which one is the fake one. But if I asked you to write down some rules to tell me why that's the case it would be hard to do in a past three of you to do it. I bet you give me completely different answers and if if if if I ventured you three more pictures one of us would fake was fake probably what you said would not apply those three talking to learn somehow and an algorithm the

same kind of ability that we already have is humans to recognize this one that's different reason for this is it's one of the areas we've been using deep learning. We apply with an auto in particular type of neural network. Where will we give it is not pictures of people's faces, but the behavioral versions of that I would talk to meet lots of things about user Behavior over some. Of time and we do this for lots and lots of users and we assume that the Dirty that behavior is normal cuz that's mostly within our data if there's some weird stuff in it, that's okay. And we use that to train a

particular neural network and the structure of this a kind of force. Is it to compress all that input into a smaller representation kind of like a compression algorithm and then you say what can we recover from the original data and it has to learn the right way to represent the things that best describe all these normal users and it it doesn't care about the fact that it can't describe other things. And then when you handed a new piece of behavior for somebody who's doing something strange, we can actually measure that degree of anomaly in the output of this neural network. And so it

kind of gives us a way to kind of find these surprising surprising types of anomaly to it down a bunch of this and when we do it and compare it to users on which we run our other indicators we found is that often times as overlap only ones that are anomalous by this particular algorithm are ones that we eat for which we detected some of these other indicators, but sometimes it's not enough cases where it's not sometimes when you go and let you go like that really was weird, but probably not security relevant, but other times it looks like it may have been So this is complementary to our

existing ones and we're looking at now is what's the right way to expose Us in the product to give you a different view of these other sorts of things that are at least worth looking at that you can then go in and look at the events do the search isn't that investigation to decide for yourself for this user really is doing something that that wants some attention. So real quick before I head back to Matt Two more slides, but one thing is what we don't have a lot of today is the ability to learn about what actually did result in a security incident or or more to the point what things that we've

put in front of you as an admin looking at at our output where you agree and where you don't agree about these indicators telling us there's something delicious is going on. So we have this. I think I only have one in Decatur now, but we are now a place the way that we can have you report false positives where you basically say, yeah this location really was trained or even more so yes, this location was strange and it was a security incident or it wasn't Tokyo first kind of a large or feedback loop ourselves to help improve the models. Alexa, everything I've talked about it with respect

that users and user risks, but there's more than just users out there you have other assets you have the other other entities for which there may be unusual risky behaviors that you want to understand and monitor and remediate and one we have in the product today is risky shares again with content collaboration. So if you share a file we monitor what happens with that share link and things happen like an anonymous user download file through a share link that was a sensitive file will amplify some risk for that and there's a whole portion of the product that gives you visibility into that

allows you to take actions on that share link as well. So without an attorney fact in that take it to the top. Awesome. Thanks, Jim. Know the velocity features. I'm going to talk about other last set of Concepts that I'm going to talk about are the concepts that you will use in the product features that you can configure the UI. So we're going to look at Services actions and search Services often times. When you listen to product managers talk about pricing packaging you're going to use the word we use the word services but

we often tend to use those words. It'll interchangeably with itself is a framework of platform and then you have Services baked into it. So you have security right at the top left corner of my screen you have security performance an operation to do a web applications that are being built on top of it the Citrix analytics platform. And the platform itself provides all the services like in our big data governance and so on and so forth. So this applications

connect to the platform using an API Gateway. So if you want to call with one of us and someone says staying with service in Citrix analytics to use setting to the services in the top left corner, so basically directory services today within subjects analytics and then we're going to add more as we go this up is because Ashley start extending the platform, we will start providing, you know, apis for third-party integration. We're also going to provide other ways for in the US customers to get the data out Saturday. There is a way for you to

connect your skin to Citrix analytics and consume our data we do that to Costco, but there will be Northbound apis provided as well. It's the whole idea is I think what I want you all to take away from this particular slide is that Central Time Athletics is a platform with Services built on top of it and we can integrate with third parties. As well as other products east to west and north to south as well. the second key concept and it's something that I think differentiates us from a lot of other analytics vendors is the ability to take action only user

for now, but then on any other profile entity that lives within the service So there are two kinds of actions you can take mine relaxants. So today in security analytics you can go to user's profile. You can take action on the user. You can start session recording you. Can you log off the user you can lock the device and things like that. So we called Emmanuel action, but in addition to this to take action on our use us, so how do actions look so I'm going to show you how the how the manual action works but policy-based actions was the same way. So when you are in the user profile over here

Diamond Georgina Caesar profile, you see her doing something suspicious you want to take action on Georgina, what you got to do is Pick actions and pick an action in this case. I'm thinking of, you know, Idaho to watch this right and I click apply. When I click apply what really happens if an action field is created action field is created and put on a service bus service buses. Just another Rodger component recreate the payload put it on the service bus and the citric

products full from the service but they've been configured to do so 7-Day policy action from the service from the service pass the exam of the action payload and they decide on whether or not they can take the action which is the Citrix product that is sending data to Citrix analytics the status message back telling us further action succeeded or failed but I talked about actions to customers customers assume that there is some kind of a direct connection between the product. What is a citrus

product and Citrix Analytics? Products that is complete isolation between the product and the subjects way between subjects analytics and the underlying Citrix product communicate through the service bus gets placement service but gets consumed by the subjects product and after consumption the stage of support comes back to us. The third concept is self-service search Jim spent all this time talking about machine learning and how we abstract all the raw data from the user

the effort of going through all the raw data. We we save you from dashboard fatigue the machine learning algorithms take care of analyzing the data and showing you alert but matter to you if you want to investigate if you have to investigate an incident, but if you want to further troubleshoot a user's performance problems in performance analytics you need access to that data is our access is your access to Raw data. Try it. So when you turn on analytics you start

doing things in the workspace app, where you go to content collaboration do something that you know Shadow file download a file when you as a user. End-user start performing action even start looking into analytics and those events all show up all the Revenge show up in search the search is available in the top right corner of the analytics window. Not sure you can draw on different kinds of fairy so you can draw and you know a Boolean credit right here in the window, or you can do a faceted search. Or you can you can also

toggle between the different subjects data sources, if you configured multiple data sources, you're going to see date of birthday the source and the last but not the least you can export all of these results to CSV asked me to say do I have access to my doll data? Can I go into a database and pull it out? As of today? We don't provide you with that capability. But what you can do is exported to CSV and then visualize it in any to love your choice, but that said if a security expert

export your processed event to be processed events, like Risk indicators user risks course user profile to a Sim Saturday. We support Splunk so you can send all of that to Splunk and then combine that data with all other day that you haven't flunked and paid your dashboard in future we plan to Support more Sims. But as of today, we do supports dimm vs. Splunk you have access to that you can search that all day that you can quit either all data and most importantly going forward. We going to have custom

search and customer reports. So if you want to play it in a weekly reports that they want to send you a management. This would be where you get the speaker. And that's pretty much and I'm glad to say that pretty much done and you have just one more session to go before you go to the party this evening. But before you go just remember the next session that I highly recommend this send to a one which is performance analytics. It was a sold-out show yesterday. If I don't do we have 350 people. We don't have 350 here but performance definitely was

right after the session and T-street most important got to go to feedback every time I come on CW coming to Synergy for the 80s. Thanks to all of you. Please. Give us feedback. It's the to feedback that make decisions better. That's pretty much it from both of us. Thank you, and any questions, we will be here.

Cackle comments for the website

Buy this talk

Access to the talk “Citrix Synergy TV - SYN202 - Under the hood with Citrix Analytics”
Available
In cart
Free
Free
Free
Free
Free
Free

Access to all the recordings of the event

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

Interested in topic “Software development”?

You might be interested in videos from this event

September 28, 2018
Moscow
16
128
app store, apps, development, google play, mobile, soft

Buy this video

Video

Access to the talk “Citrix Synergy TV - SYN202 - Under the hood with Citrix Analytics”
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
523 conferences
20459 speakers
7482 hours of content