Devise your Hybrid IT Performance & Monitoring Strategy
Transcript
>> Hello, everyone, thank you for joining us. Our topic today is, Hybrid IT Performance Monitoring Strategy. My name is Chia-Chee Kuan with Uila. Our agenda today is-- Will be started with a brief introduction about Uila, and then very quickly dive into the three phases that I'm going to talk about today. That is about Hybrid IT Performance Monitoring Strategy, started out with the planning phase of migration to the hybrid IT environment and then followed by, during migration what should be your strategy and continuation from your planning phase and finally, after the migration fully into hybrid IT.
What are the strategy that you should devise, to have a great performance monitoring solution. And lastly, we will do a very quick demo to see what we pitched today can be really realized and materialize in today's enterprise data center for hybrid IT. And lastly, we'll close it with the question and answer. Excellent, so some quick facts about Uila. (mumbles) Uila certainly, we are a startup company based in Silicon Valley, California. We are about seven years or so, serving enterprise IT organization, help them, performance monitoring on their data center, whether it is in hybrid cloud, private cloud or public cloud purely.
Our solution basically contains our core technology started out with deep packet inspection, by using our virtual network traffic tap, we are able to deploy the software based solution and quickly do deep packet inspection. Understand the network in your private public or hybrid environment and more importantly is for the network intelligence and knowledge through deep packet inspection to build out the application level visibility from application dependency mapping, application transaction analysis, and also integrate our solution with your private cloud virtualization technology or public cloud infrastructure server resource information.
And in the end provide a full stack application centric monitoring and performance management solution serving the IT world around the globe. So, so much about our company and let's dive into our topic today. So first thing that I want to mention is that today's topic and the content, a lot of it is devised from a survey that we did about a year ago with VMware vExpert communities. These are the people who you know many years ago are managing purely just private cloud environment using the VMware virtualization technology which you know, most of the enterprise today, you know, on the private cloud environment are, you know, virtualized, 95 or 100 %.
So these are the folks who actually took their data center used to be in private cloud into the hybrid or public cloud environments. So we did a survey within to learn from their experience that they did actually hands on for the last few years, and share that with you folks today. And certainly, there are a lot of performance monitoring strategy and failure and pitfalls that they experienced including what you see on the screen, you know, started out from the planning phase, just not knowing the current data center to the migration phase being continuously surprised and then not validating, you know, step by step and finally, after they are fully in a hybrid environment, you know, with the hybrid silo, what have they learned, you know, from what should be done and what could be avoided through that entire experience.
com. com, so feel free to download a copy, if you like today's session that may be of interest to you to learn from their experiences and recommendation about hybrid IT. Okay, so first thing that we're going to talk about today is we strongly believe to devise a hybrid IT performance monitoring strategy must start from before you actually go into hybrid IT, before your migration with a strong planning phase in understanding your current data center situation. So, you know, leveraging the survey that we did with the vExpert community, you know, we did ask the vExpert community, what are the top three concerns you know, when you move your data center to a hybrid environment?
And the answer come back with three things that's most significant to us. Number one is, I don't even know what application or services run in my data center. Many of them are managing the virtualized environment, they may know I got 200 or 3000 virtual machine plus physical server, many many application servers, but what service do they deliver and more so secondly, how the server interact with each other. I may have 3000 applications servers, virtual or physical but I deliver 200 services. So apparently, all the servers work together to deliver you know, single or a collection of services, how do they actually interconnect to have the interdependency.
The Expert actually have, you know, grayed out and then grayed black holes about that particular piece of knowledge and surely with that, you know what that would be inviting them to be-- The third concern would be as they move to the hybrid cloud, how should I design my security? So security naturally is number three, or shouldn't say, number three, probably number one concern here for the community. The next survey that we asked them is, what are the most time consuming parts of your hybrid cloud migration?
And then you probably can already guess. They didn't know the application, they didn't know the intern dependency that actually cost them a lot of time to figure out. Some of them are using tools, some of them are actually just using pencil and paper to interview the application owner, interview the IT staff to build that application level knowledge to build that interdependency, and naturally that's very time consuming. And what they have told us is that by the time they are done figuring out the inter dependency the dependency has already changed.
I should also say, the survey also asked them the question about how long do they usually take to migrate a single applications and the answer comebacks is not days is not weeks is actually, you know, over month time period. So certainly over that time, months, time of a few months or a month, the dependency can surely change. So the time consuming part is again, application ,discovery, dependency and next is architecting in hybrid cloud based on that knowledge about the application these are the most time consuming part.
A third question about the survey, I do want to share is, what do they see as most critical to them? Again, very consistently, application discovery, application dependency, is something that deem very very important for successful migration to the hybrid destination. But they also mentioned for the data center they are managing now, it is critically important for them to know the current performance criteria. How fast, how slow is my application response time before I even move a single bit of my application to the cloud, as well as what are my end user experiences?
Are they experiences five millisecond of VRP application? Or are they experienced 20 milliseconds of delay, you know, from their home or from their remote office or my healthcare application, that type of thing, they do want to have a quantified answer so that during and after migration, you can have a reference point about how much you have improved the situation in terms of service level. So then those are all interesting and important critical points for us to pay attention too. So, to summarize the migration planning, the first phase of the IT performance monitoring strategy is that all these things can be very complicated can be very time consuming, and they all suggested that you should use tool, don't use pencil and paper don't do just interview process asking people how they build it, you know, five, 10, or even longer time ago.
Use tool to get this done instantaneously or at least fast to build up and discover your application, build out the discovery, build up the dependency mapping, so you can have a successful migration as well as create a baseline for your application performance and user experience and then also, what is your server infrastructure resource consumption basis, right? At the bottom of the screen, is kind of an embodiment of some of the tool actually, this is how our tool-- How it would be able to automatically create this dependency for you for particular applications.
So this is one examples. And then so much about, you know, planning for migration. And then finally, sooner or later, you have to start migrating. So, the survey question that we asked for the community is, how and what is your approach in terms of migrating your application and interestingly, you know, 21% or so, of the staff, they actually do this what we call the lift and shift or BIG BANG migration, overnight they move all the servers involving an application to the cloud or to the final destination.
But you know, three quarters of them or 80% of them are actually doing the iterative approach that is step by step carefully, you know, move one server or one piece at a time. So that is probably something that I would personally recommend and probably is a more prudent way of doing this, to move it step by step, and surely it can be done that way the majority of the folks aren't doing it that way. So, if you choose to do it one way or the other, what are the important thing to pay attention to?
The survey tells us is that to have a successful migration is to validate as you move. Migrate with confidence, you move a step you see it happen, as you see on the screen, this is a similar application dependency mapping, as you can see when you are purely private, this is the face where you are actually moving already, and you see the bottom of the screen you have two, you know, red ball, these are in the public cloud section. As you move them, you want to see them move, you want to see them talk with linkage and dependency, whatever linkage and dependency in communication, you expect to see happen, you do want to see that happen, like the communication from the private cloud, all the way to the public cloud and from the public cloud here are the two workloads or two virtual machines as you deploy, they do talk back, you want to make sure they are talking back in the way in architect you design them to.
And whatever you didn't design to have happened, you don't want to have seen the line that is pointing back to an unexpected recipient that creates extra traffic causing surprises or even failed the deployment. So, move as you go or migrate and validate is something that the recommendation is. And during the migration, not only you see what should happen is happening, you also pay attention to you know, the performance bottleneck. Are we responding slower, are the networking round trip time getting slower application response time due to the allocated (mumbles) resources are limited, are they responding slower?
All these should be monitored along the process. So actually, you know, you may have the monitoring solution for your private cloud as you start to move the first piece of your infrastructure or application into the cloud. Your performance monitoring solution today for your private cloud is already and should have been exercise to see if it can see the hybrid environment because you are going to need it when you move the first piece of your server application, okay? And what happens after you migrate everything to the cloud, right?
Or to the hybrid environment? What are the challenges? So you know, one interesting question that we asked the vExpert community is, have you been embarrassed or have you seen anyone being embarrassed that they wrote over to the cloud and then have to rollback? 8% of them have personally experienced the rollback due to failure or they have known their colleagues having to rollback from a hybrid cloud or public cloud migration. And then we ask them, what are the reason for the failure?
And then in the end, there are two things that jumps out at us. Number one is, we move it over, it was seemingly running fine the beginning and finally, the application was slow down or not working, and troubleshooting or restoring application performance to the previous acceptable level they cannot achieve that. Therefore, they have to rollback to the private cloud to start over again redesign or find out the root cause, but in the meanwhile, they have to rollback to continue that critical service. The other reason for them to rollback is they just hit some surprises, the things that they didn't know and then hit them, you know, out of the blue.
So that's, you know, two major reason for them to have to rollback. So in the end, in the hybrid environment, troubleshooting is inevitable monitoring strategy and performance management strategy would include, you know, all the components that you see on the screen. You have your networking team, you have your application team, you have your application runs on hybrid cloud, you have your server infrastructure, you also have your cloud. And when application don't work or slow down, it could be in any place. That's the challenge.
Before you migrate and get involved with hybrid IT everything's pure private on premise, as you get into Cloud it's yet another silo. So this gets into the silo complexity not only just the team, you may have a cloud operations team, but also the tools, maybe you would acquire another tool to manage your cloud. So the silo team and silo tool, actually cost troubleshooting and performance management even more issues. What we're seeing on the market and in the real world, is, you know, when performance issue or crisis happen, the war room situation will usually involve multiple team and multiple tools.
Everybody's dashboard from the tool from their silo tool are agreeing, except when they do compare notes, they couldn't find out why the application performance is running slow. So you just run time and time and time couldn't resolve and find the root cause the stress is high. And in the meanwhile, the application is not working or very slow, causing enterprise money in employee productivity or production capacity, or either revenue lost due to eCommerce not functioning, things like that. So it's pretty painful and usually how people deal with that, you know, in that situation, they hire IT person , hire another guy because they feel that they need more skill sets to solve this type of issue or they buy more silo tools to cover some visibility holes, or they kind of close their eyes across the finger hope that the problem will just disappear tomorrow sometime it does, except it'll just come back two months later, right?
Or the most common one is many of the enterprise that we experienced, they just buy more server resources just made the server bigger and bigger or in the cloud, just buy more from the call service provider thinking the problem will just go away. And many of these actually don't work and then actually do not work in many many cases. We think there's a better way to do this. That is something that we advocate for is something that we call the hybrid IT performance monitoring with full stack visibility for multi cloud environment.
So the idea is that you have your application that runs on multi cloud environment, your IT staff, your networking staff or your server infrastructure, SysAdmin folks, they need to have application visibility, because application is what they support. They're not the application developer, yet they do need to have that visibility into the application or to the application to fully support it. And they need to be able to see each other's data in one single tool, one single pane of glass everything correlated together. So in the next kind of graphic is kind of how we see the tool, the silo should be consolidated and correlated in a way where the application visibility needs to be provided to the IT staff, to the NetOps, to the SysAdmin and where the you know, IT staff may definitely will need to see each other's data so they can correlate across the cloud boundary and compare data to have a full stack correlated view for better monitoring, for performance optimization, as well as for performance troubleshooting.
So this is kind of what we believe, and actually deliver a solution that we believe this can be very quickly and easily achieved and we'd be helping you know, the community and the market with our solution to achieve this type of strategy and this type of a vision. So with that said, you know, to back up what my claim is for this type of monitoring strategy, I would like to give you a very, very quick demo, to just give you a taste of you know, what can be delivered and how we deliver it.
Before the live demo, I just want to mention a few things about what's happening behind the scene when you see the demo. Number one is our technology as I mentioned early on, does use deep packet inspection but we do use virtual network traffic tap. These are virtual machine, virtual taps that get our hands on to every single packet. So we do deep packet analysis, of these packets flowing between your virtual machine from virtual machine to physical server, from physical server to networking devices, or even all the way to your cloud workloads.
We get our hands on those packets through the packet inspection, as well as we do, also application discovery because we deep packet inspection into the application layer. And because we can discover application, understand and classify 3000 plus application, we measure their application and response time we measure their transaction and more importantly, we build that application dependency mapping in a very distributed way. And then also, our other technology is not just about networking and not just about application, it is also about your server infrastructure.
The CPU, the memory, and the storage system, all very important for your hybrid IT whether they are sitting in your private cloud, or in AWS or Google Cloud environment, these are all resources that you pay dearly for, and we do integrate with them. So can correlate, and we do believe you need to correlate application response time we saw those server infrastructure components as well, right now only those costs money, but also they impact the performance. Therefore, our claim is that the full stack hybrid IT correlation and troubleshooting can be done in such a fashion that is most efficient for the hybrid IT strategy.
So having said that, I'll just show a few very quick screenshots or a live running of our solution. So what you see on the screen is what we automatically discovered, application dependency mapping. This is one single service, but as you see on the screen, it's already consisting of, you know, maybe 10 application servers already. And not only you see the inter dependency, what service do they serve, you know, how are their application response time? This is not just a pointing time but you build up with a series of time will generate reporting for all this type of information as well as for this application dependency mapping.
We also generate CSV file the spreadsheet, so you can see not only who's talking to who? What application they are? What ports, TCP or UDP ports that they are using? So when you're designing your networking kind of firewall for security purposes, you all know what to block what to let through, or so how much traffic volumes are going through between all these links, right? All these traffic volume, if you were to move anything to the cloud the traffic over here definitely would be of great concern because you may have to pay for it once you're in the cloud.
So all these detailed information, you know, can be pulled out very quickly by using our solution. Typically, our solution deployment is about an hour or two for, you know, hundreds, if not thousands of application workloads. And then generating this kind of graph is just, you know, a few minutes, you can automatically discover all these, right? So this is before you migrate to the cloud. You have all these and also you will have the baseline for the performance of any application, so you know, what your baseline is, and also, you know, the end user experience.
And also during migration. During migration, as you migrate, you know, as a different time, our claim and our recommendation is during migration, you want to see it as you migrate. And what you will see during migration is this type of thing as you move to different places, as you go moving time, you can start to see your workload showed up in the cloud. Your private Cloud environment continue to be there, and then all these things you're seeing these communication and workloads communicate across the cloud boundary.
This should be something that you expect them to happen, for example, this one, you can mouse over to it. This is the SQL a transaction or SQL connection, you know, going through how much traffic volume it is, how fast is the response time, all the way down to when something don't work or starting to slow down. It needs to be quickly correlatable into what is slow? (mumbles)be slow. Why is this slow? Clicking on HTTP for quick troubleshooting? HTTP application response time, CPU is under provisioned, back end server, my SQL services, three of them are all responding good.
Storage system is little slow, CPU is really slow, click into CPU information, to look at all the CPU information to help you troubleshoot really quickly down to a process level. All these are, you know pretty easy and clickable, to have everything in your fingertip during migration, and even after migration and troubleshooting is using the same tool, where you can see what is going on, you know, in any particular server and to whatever server and then this is not just about application or networking, right?
This all the things you see can be correlated into CPU memory storage from the application level down, or the full stack nature of what we pitch is from the infrastructure level up as well. Meaning in your hybrid IT, you want your IT folks be able to-- Your NetOps guys be able to see the networking, see the networking data, analyzing every single packet. So certainly networking is all (mumbles) how the flow is, but more importantly is when you're looking at networking, you definitely want them to have the application visibility searching here, you can see the application is HTTP, my SQL and some of the TCP transaction in here as well.
So these are, you know, the visibility that definitely need and tons and tons of networking analysis in one single pane of glass, right? I don't have too much time, I probably need to wrap up fast, but I want you to get the idea that it's not just application down but also from the infrastructure up into application as well. Networking is example CPU and analysis for your hybrid cloud environment. Whether this is your on prem database or cloud you want to be able to see it for memory, for memory analysis, where the hotspot is as well as storage system performance.
And the key is always you see the hotspot wherever they are and also you see what application it is such as this virtual machine, is storage system is slow. It is this application correlation wise you correlate quickly from the application response time, quickly to CPU memory storage and back end services for troubleshooting purposes. So these are all very quick and easily done. And then before I adjourn the session, I do want to have one quick call for action, that is, we're offering you know, we want to pitch this entire demo to you have a great introduction meeting with you.
Please sign up with us for a quick intro demo session to thank you for your time we're offering $100 on Amazon gift card if you're willing to spend 45 minutes with us. com, I'm sure you'll find the time very well spent with us if you sign up with us. With that, I thank you for spending the last 30 minutes with me and hopefully this is meaningful and useful to you, to devise your Hybrid IT Performance Monitoring Strategy. Thank you very much. (upbeat music)
