Explain performance capacity planning

- So you've been tasked to ensure that an application or a business function performs within what is necessary. How do we do that? Stick around as that conversation starts right now. - [Narrator] You are watching "ITPro TV." (upbeat music) (graphics whooshing) - Hello ladies and gentlemen, thank you for tuning into more of the CompTIA Cloud Plus. In this episode we're going to be diving into a little bit more of that planning. Remember, if you fail to plan, then you plan to fail. And if we're deploying applications, if we're deploying some kind of software, some function, some server, we need to make sure that it performs adequately and according to what the business needs are. Ronnie, I know this it can be a very tricky concept to get right, but that's why we're in the planning stage so we don't have to come back to the drawing board after we've already deployed. Where do we get started? - Yeah, so the hard thing here is that to understand in terms of what a business actually has, we have to actually think, like we did in previous episode, we have to talk about the ideas of business capacity planning. In other words, it does everything fit the business that we actually are doing right now and that we actually need in the future. But the other one is performance as well. So it's easy to let one side kind of rest heavy and the other side to kind of take a hit. But what we really need to balance of both, we need to actually take the idea here of planning for performance as well as for business. So let's make sure that we actually use a couple of examples to kind of help us out. So Wes, when it comes down to the idea of performance capacity planning, I always like to think about the idea of rental car here. - Mm-hmm. - Okay Wes, I've been stuck with one of the economy rental cars before in a hilly area of the country where I started to go uphill and I realized something, I was standing on the actual gas pedal and it was seems like it was actually getting slower. Even though I was in low gear, it was still getting slower. So Wes in that sense, the performance of that vehicle itself was probably not what I needed in the hilly parts of the country. - So if you were going through the rolling hills of New Hampshire or maybe planning on climbing the Mount Washington there- - Yeah. - You probably should have planned for whatever the vehicle is that's going to take you there has a little bit more performance under the hood. - Yeah. Not the Ford Festiva- - Not going backwards in time. - That I ended up with there. - Yeah, not going back in that. So in other words, planning for that would've actually been helpful if I- - Sure. - Actually understood what the terrain was like. - Mm-hmm. - I wasn't even that high. I was just in Virginia. - Mm-hmm. - So when you had that, you actually think about that. The other one, Wes, is probably something we've all experienced at some point, which is you have a moving truck and you go, this is the right size moving truck for us. And then you drive it home and you load up every single box to the very end and you still need another trailer. - I could see that. - I couldn't get the rest of it in. - That most definitely. And I could also see that, you know, we've talked about other topics like for instance, CapEx, right? Where you've already spent something and you don't get to actually utilize all of it 'cause you don't need it. That's a lot of wasted money. I can see coming home with a big truck, that you put what you thought was all of your stuff and you still have a lot of room left over, so you maybe have wasted money as well. - Right, so yeah- - Planning is key. - Yeah, it's actually very key for us to do so. - Mm-hmm. - So you might think, "Well, Ronnie, they have websites that actually help you to plan how many rooms that you have and do that." Well this was before they actually had those websites. That's me. So, I was like, "Oh, yeah-" - Well, the good thing is- - Of course, I've-" - A lot of cloud providers- - Right. - They do provide some kind of guidance, right? - Sure, yeah. - In their documentation- - And that helps. - Yes, definitely. - And of course another one I like in terms of actually planning out the things here is you might actually have think something like this, which we have actually done before Wes, even before "ITPro TV" and the company we were working for before new building built out and it's like plenty of power outlets in the rooms. - (laughs) That's right. - And then you get there and it's like, yeah, there's two, there's two actual outlets in the room and you're kind of like wondering like how is this going to end up working? So it doesn't work as successfully as we think. It's like, hey, maybe we needed like 10 more power outlets actually in the room itself. So having things like enough power is always key when it comes down to, to what we're thinking about. So the same thing when it comes down to performance capacity planning as well. Performance is a key aspect of everything that we do today, right? The idea of actually waiting for something to actually work and develop. It's the frustrating part about the cloud that's is really frustrating for service technicians Wes- - Mm-hmm. - Because it takes time for the commands to actually end up getting the convergence that we need in the cloud. When you're doing something like service help desk, a lot of times, and I know I'm kind of diverging here for this for a moment, but when you're doing something like that and you tell somebody like, "Hey, here's what you're going to do," and they do it and they're like, "All right, it's not working." And then you might have to tell 'em like if you'll wait, you know, it might actually only be 30 seconds, but usually I say if you'll wait, let's say 10 minutes, everything will settle down because all that has to kind of, you know, converge out there and do everything that it needs to. People don't like the idea of waiting. So when we start talking about performance requirements, then, okay, when it's actually just our cloud and what we're doing here are the things that actually matter in terms of impact for performance. Now Wes, we've also of course been through a plus before we know those four areas, right? CPU, memory, storage, networking, those are kind of main things that actually impact overall performance for a regular computer. Well in the cloud, those are things that we also need to consider as well, okay? And the whole idea behind this is to actually have the appropriate CPU, the right amount of memory, the right number of nicks that we actually need, the enough storage that we need here. And ultimately the only thing that the end user cares about, Wes, a lot of times when it comes down to cloud is why isn't it working as fast as I think it's going to work? - Right. - Okay, so when it comes down to it, you hate to say it, but performance is absolutely key. If somebody thinks things are actually going too slow, you can imagine how your customers, you know, that you're actually trying to service in the cloud how they think it's going as well. So we want to ensure that when we start planning that it does actually feel like what we actually expect it to feel like that it is working that way. And everybody can actually say, "Oh no, it's about making sure that you right size this or that." It's more than anything, right? About how your users are going to feel when they're working on the network itself. So those response times have to actually feel like that they're snappy and they work. - Absolutely. And that's why, you know, that brings up another topic that we're going to talk on when we talk later about testing and stuff. There's a big difference between functionality testing- - Mm-hmm. - And usability testing. Right, when we talk about functionality testing, does it work the way- Does it do what it's supposed to? Usability testing's a little bit more about that customer experience. And that customer experience at the end of the day is going to be the one that you're going to hear the most about a lot of times. - There's no doubt. All right, so now that we understand those particular requirements and the services that we're going to actually take a look at, we also have to actually think about the idea of some considerations too. So what is it that affects all those things that we were talking about? Well, it comes down to what we call system load, okay? System load for us gives us this chance in terms of planning to actually ask the question, how is it that every single resource in our cloud environment, what is its utilization like currently? Okay? And the reason why we need to know that is does somebody actually have a bad experience with it or is everybody feeling like it's actually okay? And then we have to also determine like is it because it's the less, you know, there's only four users actually using it versus the 99 that's normally using it as well. Also, when we do this too, we might actually find out are we currently at the limits of what our systems can handle? So we take a look at CPU networking storage again, are we actually at the upper limits? If we were to add in another one, it'd be like that straw that broke the camel's back just too much. And even though we know that we can spin up more, it's still one of those things where we want to plan for it and not just go, oh, I didn't expect that to happen. 'cause that actually seems like you really didn't know what it is that you were actually trying to do here. And are we actually at that point where we just need more just because of where we are? That should never be a surprise thing, Wes, okay? Is that, oh yeah, I knew we were going to hit this limit two years ago and yeah, we just haven't upgraded here. It shouldn't be where it's like what we hit our limit. I mean, that should not be the process, right? This should be something that we would've planned out and said, here's what we're going to do when we get to this limit here. You know, whatever, it's 75% utilization of this means we're going to spin up another device. And there's some things that are automated to help us out to be able to do that. But at the same time, it's knowing the idea that you can plan for performance and know exactly what level you want to jump up to. Because Wes, if you're like me and you're going, you know, look, we're at a hundred percent capacity. Well I'll just double the capacity right away and you know, we're often ready to go. And now the bill also doubled and- - [Wes] That's right. You've overshot the runway. - Yeah, I've kind of overshot a little bit more. So great performance, but at the same time, maybe a little bit over two. User density, the current number of users versus also the max number of users. Those are all key. Now why do we really care? Okay, the reason why we actually care about this is because of what the future holds. We can't expect things to be maintainable. In other words, it's not like it used to be Wes, where we'd go, all right, this should be fine for the next three years. - Mm-hmm. - The amount of data that's grown in three years time today, Wes, versus when we first started in the industry. It's insane. Okay? When we start talking about, so Wes, we can even use an "ITPro TV" example here, okay, started "ITPro TV" about seven years ago. That first year, I think we might have come up with, let's say five terabytes of data- - Sure. - In terms of video. Wes today and just- - [Wes] Hundreds of terabytes. - Hundreds of terabytes. Hundreds of terabytes of raw video footage that we have. That's all there in the cloud for us. And you know, you asked us several years ago, I was like, we might get up to to 20 or 30, you know, terabytes maybe. But now it is we're hitting hundreds of terabytes of data. So the idea of load here is actually very key. So we can take what we've actually learned now and from our past performance to help us to actually do better performance planning for the future. So how do we do that? We actually start working with some type of trend analysis to help us out. Wes, we have to begin with what we always need to begin with, which is a baseline. - Yeah, that's right. - Okay? So Wes, how important are baselines? - The baselines establish what your basic level of performance. The acceptable level of performance that any system should give. And it's going to be useful in a lot of things. Not only trend analysis, but future monitoring. And are we actually utilizing what it is, if you will, that we've deployed. And the other thing about baselines is it, it gives us a consistency. And that's really good too. 'cause once we're planning, we've got that established plan, we want to make sure that everything, any resource that we deploy, adheres to this pre-configured, you know, recognized type of configuration. And that's where the baselines really come in because ultimately they can be used to track back and see if we're having some of these other things. Are we identifying patterns? And I know we'll talk about things like anomalies coming in as well. - Right. Yeah, so when it comes down to it, baselines are for us a point of comparison that we can have. - That's right. - Hopefully it's an honest point of comparison that we have. So that way we can start trending and looking for the idea of patterns, like what West was mentioning here. The idea of understanding historical data that we can keep and we can monitor, might be able to tell us in the future how many users we're expecting on our network. Okay? So for example, like when "ITPro TV" first started, we had what, five people when it comes down to it, well, today we have over 70. And then with the, the idea of ACI also coming on board, we're now 150 to 200 people- - Mm-hmm. - On board here. And so now the future number of users, right? We can actually predict that by looking at historical data and then trying to trend what the future brings us. Because we might be able to see patterns that are actually there. Also, of course, what about the use of resources? We might find out things that we were using two years ago, Wes, we don't use anymore, okay? But yet we still had it on the background because we just never turned it off. What we might find that we actually need to increase resources in one place versus another because of that. So all that's actually key too. - You know, Ronnie, one of the things I like about the baselines is, you know, they talk about, you know, hey, let's establish the averages. Well, how do you define what the averages are? How do you define what anomalies are, right? Well, we have to know what is the established level of performance to begin with, and that's where the baselines come in. Then we can see if there's any a deviation from what our baselines is. Because as Lonnie's mentioned, things we innovate so fast and we can adopt new technology so fast. Is it important to understand where do we stand right now and how can we define things that, or identify things I should say, that kind of work outside of those averages that we're expecting, - Right, and part of that of course, is also when we actually identify those patterns when we start talking about it, is if we were to try and go on with what we have now, would that be sufficient in the future? And then where do we actually begin to apply more? So Wes you talked about the idea of anomalies here, let's talk about what that actually means for us. - Sure. - It's not a pattern. That's what it essentially means- - Right. - It's not something we can really base any future decision or any planning off of. - Mm-hmm. - These are outliers to our performance for whatever reason- - Sure. Sure. - That they might be, that ends up happening. So you will actually find anomalies and the key is to actually be able to identify them as being anomalies as well that will actually help us in terms of performance. Now Wes, you have a demonstration for us that will actually help us to kind of take a look at the idea of what we're talking about here, right? - Most definitely, and I'm using the same platform that we used in some of the previous demonstrations. And in fact, ladies and gentlemen, we're going to put these step-by-step that Ronnie and I have used to demonstrate this. We'll put those in the learner resources and that you can use them and you can recreate these. However, we always want you to keep in mind that remember, you are responsible for anything that you do within a cloud-based platform. So just be careful of what you're doing here because it does cost money. One of the things that I want to focus in on here is the ability to create virtual machines. Just like we've done before. It's one of the easier types of processes that if you're getting used to this for the first time, that we can kind of talk about. Now, in a previous demonstration we kind of looked over some of these things, but now I want to go and focus more on the performance side of, let's say, when we create a virtual machine. And to do that, I know I kind of just clicked on here, let me be a little bit slower on that and show you, I'm here at the Microsoft Azure dashboard and what I could create a resource right here, but I'm going to choose virtual machine. And when I choose virtual machine up in the upper left hand corner, I'm going to choose the down arrow next to create and create a virtual machine. Now, like I said, the process you've seen before. We've already done this before, but I want to focus in on some new things, some different things here as it relates to the performance considerations. Now what we're going to do is we're going to go ahead, I'm going to create a new resource group here, and we're not too worried about the group here, but just remember when you do create a cloud-based resource, a lot of times it creates a lot of additional resources that have to be factored into performance and cost as well. So I'm going to create a new cloud resource group here. I'm going to call this, we'll just call this CloudPlusVM-01. And I'm going to keep the region where it is. And the same thing with availability options. We'll talk about those coming up a little bit later, but now here's what's really important is when you talk about performance requirements, remember the generation of the virtual machines you have, because that does kind of lend itself to what available functionalities you get. And it might be something that's a performance requirement within your network. You talk about things like scalability, right? Do we have enough storage? Well, if you have a gen or one of these older generation virtual machines, you might be limited in the amount of storage you have and it might be something that you need to consider. The other one that I also want to keep in mind too, and I'm going to, by the way, I'm just going to take the Ubuntu server 2004, long-term solution gen one. And that's because it's a little bit cheaper. And I would recommend that if you're doing this at home and maybe trying to figure this out, I would do this one too, because it doesn't cost that much. And that's because I've already done the performance planning, right? I know that I don't need a lot of performance for this machine just for a basic demonstration. However, if I have an application here that needs to support a greater size of resources, I need more of the virtual CPUs, I need more memory, right? Notice that the price goes up. Of course you're paying for what you use, but we have to keep that in mind when it comes to, you know, what performance are we trying to support here? I'm going to, again, just keep the basic here. I'm going to change this. I'm not going to use a password and we're going to use our cloud plus admin as our user. And then I am going to try to type a super secret password in here the same way twice. I have to tell you though, off screen, off camera, I did not do so well last time. Oh, there we go. All right, and I'm not going to give it an inbound port. We'll talk about virtual networking coming up in later episodes, but I don't want anybody to have access across the Internet to this. Now here's the next thing. Keep it in mind just as a real quick recap, right? How big the machine is is directly related to the performance capacity that you're going to get. The generation also dictates like what is the overall size and the resources that are available to the machine too? And next is discs. Remember that one of the four that were in there, we've actually taken care of two of them, right? Memory and CPU. Ronnie mentioned also storage, right? It's not only planning for the storage size that you need, you can see some of these sizes are quite large, right? We can go all the way, go up to a two terabyte drive. And again, you have to plan for that, right? More on the performance side too though, is what kind of disc are you need to use? Now ladies and gentlemen, let's be transparent here. We're trying to be vendor neutral, but this is a very vendor specific based technologies. I don't want you to worry about the difference between premium SSD and standard SSD or standard HDD. From your a plus training, I do want to know the difference between an SSD drive and an HDD drive, right? If this is a machine that we're going to use archival data and we're going to access it maybe once a year, I could get away with a standard HDD 'cause I'm not going to infrequently accessing that information. However, if I have a high amount of IOPS, input, outputs, reads and writes that I need, I'm probably going to want to go with a premium SSD and again, a higher rated SSD, because that's the performance I'm going to need. So again, part of that consideration when it comes to performance planning. We're not going to worry about encryption, we're going to talk about that later in some security based episodes here too. But we do have that ability. Same thing with networking. Remember networking is a third part of those four resources that Ronnie mentioned here. We do have to pay attention to what kind of networking we need as well as, you know, the virtual networks creating them. Again, we're not going to worry too much about the networking side, just to understand that you do have to prepare for the performance that you might need. All right, see we can do accelerated networking and that directly again, gives you the high performance. Now, here's another thing that I think is really interesting, Ronnie. I selected a very basic VM and notice that based on that selection, I'm not going to be able to use that advanced networking, right? So I don't get that high performance. And again, that's going to be something that you have to consider when you have a business application or a business requirement or function that you need to adhere to. Now with the manage, we'll talk about this setting here coming up and again, later episodes. A lot of good stuff that we can see in here. We've already kind of talked about monitoring, nothing really here that kind of affects us- - The performance. - Too much performance. - Yeah. - For performance. Same thing with advance. We'll talk a little bit more about this. I will tell you though, here is one thing I'll let you all by all means read the difference between the VM sizes and what they give you access to, whether it's virtual memory notice OS disc size limitations, that directly affects your storage side of your performance as well. So those are more to some of those considerations. I'm just clicking next here. Tags, I don't need any tags. If I decided that I wanted to automate this approach, again, you can see the information here, what it's going to cost me here, you know, notice the validation has passed. We've got some additional information in here. Again, we're going to go kind of blur this out here. Sorry, I don't want any phone calls at the house. You can always contact me via email. (laughs) And we just kind of review, you know, the choices that we have made here, as well as the performance levels that we've needed and the various resources that they allow you to allocate here. I could download the template for automation and then we choose Create. Again, when you are creating these, make sure that you do pass validation. Notice that the deployment is in progress. Now there's one last thing, Ronnie, that I want to mention that we're going to give a resource to our users out there. And this directly goes along with a lot of the stuff that Ronnie has mentioned as far a as performance preparation, okay? Keep in mind, this is the Azure specific. We're up on Microsoft's documentation here. I don't want you to worry about what they call it as far as the sizes, but I do want you to see and tie back general purpose CPU to memory ratio balance, right? Compute, optimize. We need high performance machines, right? Memory, optimized storage. And then you can see high performance compute. Do you need something like an HPC cluster? Again, all things to keep in mind, but remember the big thing is understanding what those business requirements are will allow you to meet whatever the performance capacity is, you know, that you need to deploy. - All right, Wes, well that was actually a great little demonstration walking through, showing us how we can take what it is that we planned and literally apply them into a cloud provider for us to get the best performance that we need. Well, there are more episodes to come for Cloud Plus, so make sure you stay tuned for that. But we are done for this one. Thank you very much. Join us there. (upbeat music) - [Narrator] Thank you for watching "ITPro TV." (graphics whooshing)