Video

SolarWinds Lab Episode 55: Hybrid IT Diagnosis

Hybrid IT brings considerable new complexity and demands new tools, skills, and methods. Unfortunately, now more than 50% of IT departments of all sizes admit to not only supporting hybrid IT operations, but have been diagnosed with multiple, sometimes debilitating symptoms. Workloads split between on-premises infrastructure and cloud, multiple cloud environments, and reliance on critical but opaque SaaS services require an efficacious remedy.

SolarWinds Head Geeks Patrick Hubbard and Leon Adato dig deep into the challenges and promise of great service delivery, even as your IT processes move out of the traditional safety of your data center.

Back to Video Archive

Episode Transcript


Hello, and welcome to SolarWinds Lab. I’m Leon Adato.

And I’m Patrick Hubbard and today— actually, hold on a second. What did you guys think about that new open? We finally have a new open.

Yeah, it’s been four years, I think.

And it served well.

The part that I liked the best was how our video team sort of captured the essence of each of the geeks. You know, just really quick. It was really fun.

Yeah, but, that’s because they didn’t screw your head on with a robot.

Correct, I did not get my head screwed on by a robot, and I’m okay with that.

That’s right, and we have a new logo. Got all new lab coats. I think your other lab coat side. There you go. Yeah, and I like the green and the animated bot. Really, really cool. Again, keep your suggestions coming, and then we’ll have to have some new swag for these.

Oh, yeah, new lab coats, certainly, but other stuff too.

Danielle’s going to love the fact that we just mentioned that now. So anyway, today we want to talk about something that’s really very interesting, and that is both interesting and unfortunate, and that is hybrid IT.

Yes, and for some of us it’s more unfortunate than interesting, but it’s worth talking about. Hybrid IT is the conglomeration of IT when you have some resources on-premises and others out in the cloud. And that creates these interdependencies on both, that creates a lot of increased complexity as well.

Lots and lots of complexity. And that’s the reason that we wanted to talk to you about it–is that both in-person SWUGs and, if you look at our 2017 IT Trends Survey of admins like you, this is the first year here in 2017 that more than 50 of you admit to having the headache of hybrid IT.

50 percent of you. It’s more than 50 percent. So we’ve just finally gone over the edge and it’s actually worse than that. Of those people who have cloud workloads, the majority have workloads on two or more platforms, which creates its own interdependencies, and it’s even— each platform comes with its own peculiarities.

Enormous peculiarities. And so what we’re going to do first is define what hybrid IT is. And we’re pretty sure that you’re going to find that you’re spending an increasing amount of time managing something that was supposed to be really easy–shifting some of your workloads or storage out of the building.

Right, that whole lift and shift thing. And then we’re going to be looking at specific challenges for small, medium, and large IT operations.

With a bit of screen time on how to improve the situation.

Yeah, I think we’ll have just a little bit of that. We can also outline some of the best practices and tips to minimize the impact on hybrid IT operations.

I may even throw a word on application performance monitoring–actual APM, code injection, cool stuff.

Yes, and that was in the survey results as well.

Then finally, we should wrap up with some prognostication.

Yes, yes. Like how long is it that IT pros will have to live with hybrid IT? I see it now.

Yeah. And the answer, by the way, is forever.

Spoiler. Also, bummer.

I think what we ought to start with is a definition of what hybrid IT actually is.

[Sighs] Okay. Like I said before, hybrid IT–some of it is on-premises, some of it is in the cloud. It’s more than that. Really, what you’re talking about is that IT running the same way traditional IT always ran. Bob and Mary in the IT department used to get the call that the file server was down. Or they got the call that the inventory application, CRM application was down. And the fact is, in hybrid IT they still get the call. But the file server is Dropbox and it’s down.

Or it’s S3.

Or it’s S3, right.

And it’s down.

It’s down, yes. Or the CRM application is down, but it’s Salesforce. Or, better yet, it is actually a homegrown, custom CRM app, but it’s using micro surfaces that are running off of Azure. So that’s the problem with it–is that now the IT pro is getting those calls that it’s down but they have to fix everything.

They’re still responsible because it’s the endpoint users. It’s still the user experience issue. You’re going to get the help desk calls. Only now, you don’t have the span of control that you had before. And it’s not your fault, right? This is the result of a lot of decisions which were innocently enough made. If you are a very large brand and you’ve decided that you’re going to spend 20 million dollars or more ongoing with hybrid technologies or outright lift and shift, or application deconstructor to cloud, it doesn’t necessarily apply because you probably added systems in place, or at least procedures in place, so you think of this differently. But the added cost and complexity of using a hybrid approach is something that you’ve built into your costs. We’re talking especially— and talking to you guys at SWUG, for example—is you started with innocent enough decisions like, ‘You know what? We’re going to store our backups in S3. We’ll do that.’ Or, ‘We’re going to use a SaaS product like sales wants to use Salesforce. So fine, we’ll do that.’ But you’re going to have Active Directory integration with Salesforce, which is still coming back to your on-premises data center. And at the same time, maybe you say, ‘Hey, we experimented with AWS, RDS for SQL storage, but we really find that Azure SQL actually works better for our legacy application, to be able to sort of transfer it over.’ So now, you have two cloud providers with completely different interfaces. It’s cloud and it’s SaaS and it’s…

Redundancy and it’s high availability and it’s back, yeah.

Right. And all of it has different interfaces. And so now everyone on your team, or at least parts of your team—especially if you’re a generalist team–if each of you has similar responsibilities–or maybe you’re the only person with all of the responsibilities–you’re now expected to learn all the different monitoring and management techniques to manage each one of these technologies. So that’s what we’re talking about with hybrid IT. And the reason we talk about it, I think, is that no other vendors talk about it, because they don’t want to admit that it’s a problem because it would get in the way of them selling you something new that’s going to add to the monitoring and management burden.

Which is why I think a lot of organizations find themselves in this awkward teenage acne-ridden phase. It’s like that time that I was a freshman and I asked the homecoming queen out on a date. It was really awkward, okay. It was just really, really awkward.

You just can’t get over this. This must’ve really burned.

It was a formative moment. Yeah, it was.

So the point is that it really is this phase that you’re not quite a grown-up, you’re not quite a kid, you’re sort of in the middle. And we’re going to have to find a way to be comfortable in our own hybrid IT skin for a while, to do this. I want to play a little game. You were talking about the Active Directory.

So this is to prove to them that they are actually, probably hybrid IT.

Yes. Because there’s people out there right now who are saying, ‘No, no, no, no. We don’t have this problem right now.’ Okay, we’re calling this “You might be hybrid IT if…”

You might be hybrid IT if… You move to Office 365 and scream when you discover that you have a local mailbox server.

You might be hybrid IT if you balk at your RDS bill and expand to Azure for SQL Server.

You might be hybrid IT if Salesforce breaks when you reboot your local Active Directory controller.

Yes, and you might be hybrid IT when employees head home early because you have a WAN outage.

That actually does happen. We saw it here one time. It was pretty shocking. And you might be hybrid IT when there’s an outage and you spend more time figuring out where it’s running than fixing it. [Glass shattering]

Before we dig too far into more of the how to fix the hybrid IT issue, I’d like to take a minute and get up on my soapbox. Here’s my soapbox. And talk about “What is monitoring?”

Before you do that, what’s really interesting, this is actually a slide, and it says, “What is monitoring?” and it has the fantastic new color scheme that matches with our new open. One of the reasons that we’re actually passionate about this and wanted to do this episode focusing on hybrid IT is that we just got back from presenting this at a couple of different shows. You showed it to Bynet.

Bynet, Israel.

And I showed it at InterOp ITX. And the reaction from the folks that we presented it to— and this is broad swaths of IT—discovering that ‘Wow, you know, I really did kind of back into a lot of complexity there that I wasn’t expecting.’ And making them feel, ‘Hey, it’s okay that this is harder than you thought. And by the way, nobody bothered to really mention what the true cost of it would be.’ We are pretty highly opinionated and so if it seems like we have something to say about this, we really do.

Right. And also, the comforting moment that I found was, ‘Oh, I didn’t do it wrong. I didn’t make a mistake that I have to admit to a mistake. Everybody’s having this same situation; everyone’s having this same experience.’ Before we dig too far into that, I just want to talk a little bit about what monitoring is, just a reminder. I worked in a place—this is one of my favorite stories— I worked in a place where uptime was determined as being 100 percent minus the number of tickets they received. Yeah, that was uptime. Okay, this is not uptime. But my point is that it betrayed an attitude that monitoring was tickets and monitoring is not a ticket. It’s not a page, it’s not an email, it’s not a beep, a blinky on the screen, it’s not a poke in the shoulder or whatever—that’s not monitoring.

And if you’ve watched this show, you know that we would, in fact, argue that less alerts, for example, is actually better monitoring.

A sign of better monitoring, right.

It’s inversely proportional.

But it doesn’t still answer the question of what monitoring is. Monitoring is nothing more, nothing less, than the ongoing, regular collection of metrics and data from a set of target devices. Whether those targets are on-premises or in the cloud or in between even, that is what monitoring is. Everything else–your reports, your views and screens, your alerts, your stars and stickies, and pokes in the shoulder–all that is a happy byproduct of doing the first thing, the monitoring, correctly. So what we’re talking about here is continuing to do the same thing that we at SolarWinds have always talked about, which is doing monitoring correctly–whether it is complete cloud, whether it’s 100 percent cloud-based, or it’s 100 percent on-premises, if you’re still lucky enough to be living in that little bubble, or in this hybrid IT environment.

That’s true. But monitoring is not the same, at the same time. Not the same at the same time. Meaning that the interfaces are different. You go from easy discoverability to a lot of opacity in terms of being able to know the details on how things are running, and some of that is expectations of the provider. Whether it’s SaaS, the cloud—the point is, ‘Hey, I’m sending this off. This is somebody else’s server; this is somebody else’s problem.’ Except you would still like to have better visibility into it. It means that you’re also going to learn a lot of interfaces. Like if you’re really great on the command line, if you can get it to your Nexus and blaze away on iOS, and quickly manage all of your interfaces, policies, and everything else you can imagine, and then you go to set up a VPN on a VPC in AWS, you will find it to be somewhat different.

Right, and actually we have that experience. Recently I was sitting with Tom, Tom LaRock, and he was showing me Linux on SQL. Yes, but I meant that the other way around, SQL on Linux. Sorry, Tom.

Stop, reverse it.

Strike that, reverse it. Anyway, he was showing me SQL on Linux—which, of course, I was really excited about—he was very comfortable with, and we noticed that within, no exaggeration, 45 seconds, that instance was starting to get hammered from out of country with systems that were trying to get into it, trying to hack into that SQL instance. It wasn’t running anything, it existed for exactly 45 seconds and that was when Tom came to the realization, ‘Oh. I actually need to understand VPNs.’ And it was something he had never done before. So even if you think to yourself, ‘We’re going into hybrid IT. I’m going to manage servers. The servers are over there, they’re not over here.’ This is not just a simple ‘colo’ kind of mindset. You are going to have to learn new ideas and techniques whether they are new, actual techniques and technology or they’re old techniques like VPN. But now, they’re part of your responsibility set.

And to your point, that’s Azure. So we’ll talk about that again in a second. I was, in that first example–I was thinking, of course, because I’m kind of an AWS early person and that’s where I spend most of my time, is if you, to that point, you launch a new application. So you go set up your VPC, you set up the VPN, the endpoints, get that working. Set up your routes; get your subnet routes set up— all of these in completely different screens by the way. You can get used to it and actually, you’ll start to do this from the command line, mostly, once you start actually to use the command line tools. Or at least, ideally script it and be checking those into source control. That’s a whole other topic, but let’s say you did that for AWS and you said, ‘Great, I now have transitioned, or at least now I’m spanning both. I can do my command line config in iOS and I can go over here, and now I can navigate the peculiarities of the IM and role structure to ensure that my application is secure. Like, I’m going to limit maybe the endpoints that can access that SQL Server instance, to prevent what you were just describing.’ That’s Amazon. So now, I’m going to go over and do the same thing in Azure and I can use all the same config…

Mm-mm.

No, it’s an entirely new set of configs. So conceptually, it’s the same but the tools again are all different. And the monitoring tools for each one of these are different. And that’s that example of the pain that people are not expecting–is you’re already crunched with managing more and more resources with a smaller and smaller team, so when you go and say to your management, ‘Hey, I know you sort of pushed this on us.’ Or maybe another team said, ‘Thou shalt go to off-site application X and you’ll just figure it out.’ You come back to your manager and say, ‘Hey. By the way, I’d really like to take a training class in this. Or failing that, I need to set up a lab and I’m going to need a couple of weeks to experiment with this before I can even begin to give you close to the same security and performance or at least the ability to quickly troubleshoot or an understanding of how this works.’ Normal human learning is not built into the time budget or the education budget or a number of things. So that’s really where a lot of that cost comes from.

Right. And not only that, but what we have seen, and what we’re going to continue to see, is that you’ve got the complexity of Azure versus AWS versus…

Bluemix.

Bluemix versus whoever else.

Google.

But also, next year’s implementation of that is going to be different because they’re learning just like we are. And they say, ‘Oh customers want an easier way to (let’s say) set up VPNs.’ So now the screens that you got used to to do it now, in Azure, aren’t going to be the same ones that you see next year because you know they’re going to improve it. And improvement is good but it means also different. It also means that your processes and procedures are going to change.

Or vendors do what vendors do, which is they build shiny new things and then they go find a market for them and try to sell them. So what’ll happen is someone will say, ‘I have the greatest new administrative tool you’ve ever seen. It’s a cloud-native service. It’s hosted. It’s great. It’s all goodies and yet it’s automatable and it’s going to take care of managing your VPNs and your VPC access and rules and all the rest of it. It’s going to be great!’ Well now, you’ve just added another product. So it doesn’t matter. It may be obscuring that behind the scenes but you’re still adding yet another product. And by the way, that’ll only do–manage this one thing over here. You’re going to need another product, which is part of a related thing, or maybe an entirely separate company. So that snowball effect that we’ve had with on-premises, right? ‘Hey, we have storage. You know what? Let’s have some SSDs. Okay, let’s have a mixture of SSDs and intelligent controller. Well, now we’re going to have Flash!’ Oh great. So now, I have specialist Flash storage, right? So I’m having to learn all along as vendors are continuing to introduce new products into my data center. Well, it’s no different with cloud. So this hybrid IT problem continues to expand.

Right. Now, I want to talk about why this seems to be moving so fast. There’s a story that I’m really having a lot of fun telling. It goes back to the very first DockerCon, which is sort of where a lot of this started. And the very first DockerCon happened in the misty past of 2014–San Francisco. It was not that long ago. So back at the first DockerCon, if you were sitting there, the first speakers came out were from Docker. Makes sense, right? And they said, ‘Hey, we’ve got this amazing new technology called containers.’ And they’re really excited about it. And this is, ‘Look at what you can do with it,’ and ‘This is awesome!’ And at the end of the talk, they said, ‘Now, this is not for production. This is just something we’re playing with. We’re interested in it, here it is. The code’s out there. Play with it yourself.’ And the second speakers got up–it was five engineers from Google. And you know what they said? They said, ‘Here’s how we’re using containers in production at Google.’ Like, that was how fast the roll to production was going. It’s stuff that–was clearly labeled by the maker, by the person who produced it, as not for production–was being used. And the reason for that–Actually, I realized we have an example of that here in SolarWinds. Our web dev team, the ones who are doing all these amazing new updates, and the logos and all that stuff.

And the Customer Success Center.

And the Customer Success Center, right. They work for–they are part of the department called? You would think IT. They report up to marketing.

That’s right.

There is no connectivity.

And that’s the CMO’s budget that is now the other part of hybrid IT, which is almost this–we used the word ‘conglomeration’ before. But where hybrid IT, a lot of times now, contains elements that are completely outside of your funding path, right? So they’re not even part of the IT budget anymore, but people are still going to reach out to us. They’re still coming to you when there’s a performance issue.

Right. So the web dev team is not part of the IT structure at all. They don’t report in there. I mean obviously they know them, and they’re friends with them and all that stuff, but there’s no–And the fact is their mandate is such that any new technology that comes onto the scene, that is going to give them any kind of advantage, they’re taking it. They’re not making the same sort of technological decisions that a normal IT structure would make. But this is the environment that we are all–we, as IT admins, are all in.

Well, I think I hear you hinting at some of the scariness that we actually heard at re:Invent, last year. Which was, we have been talking–and one of the nice things about this show and our role here is that we get to straddle that world between all of you in THWACK. All a bunch of senior admins, who’ve been doing this a really long time, that we talk to on a regular basis–when we’re out in SolarWinds user groups, the THWACKcamp, the live chat for this show. But then also, we go to a lot of trade shows, and especially some of the larger ones, where we get to talk to a lot of management. Right? And so the frightening thing to me, that was different this year versus previous years, is that they were telling stories just like what you described. They were saying, ‘Hey, I’m the CFO, the COO, the CMO, and I’ve got budget. And you know I’ve just been stalled on these projects. I’ve heard about this thing called digital transformation. My peers are talking about it and it sounds cool. And you know what? I found this hotshot in my organization that said, ‘We can do all this in cloud, and hooked me up with a vendor that’s doing the integration for me. And so I carved out some budget, and you know what? I am live in six months. It’s amazing!’ And so that person, that CXO, then gets to go to their CEO and say, ‘Guess what? I did this. I did this all by myself. It’s amazing.’ And his SLA, to his CEO is not about page load times, not about outage time–It’s, ‘Were you able to implement and transform the business?’ So he can nod, and can say yes.

‘I did, and I did it in this budget, and it was understandable and finite.’ Yeah. There’s an enormous amount of glee.

They are–I don’t want to say vengeful or giddy–but there is a little bit of, kind of, ‘I didn’t have to deal with those guys in IT.’ And I had never seen that before. That and, of course, they denied that they had any challenge at all. Because if you asked them last year they would say, ‘Yeah, you know, we’re going to kind of go to the cloud, and we’re going to do it in a phased approach. But we’re nervous and risk and compliance.’ And then once then, like last year, you talk to these same people and they’re like, ‘Oh yeah! It was a piece of cake! We did the transition and it was no problem. It was amazing.’ A lot of them went up there on stage and did presentations along with the AWS crew. So the fact that they either were fibbing or had forgotten how concerned they were, or how nervous they were, and had made that transition–I think was part of that. And that’s what’s different with this. And that’s what’s different with hybrid IT. There seems to be like this emotional component on the part of executives to like force a change that they’ve been trying to make for a really long time. So a lot of times, you don’t even have control of the technologies that are being selected. And so that’s a big part of the challenge–is that normally they came to us. They say, ‘Hey, you’re the experts. I just want an app that does X.’ Or, ‘We’ve picked this packaged application but you’re the experts at storage and network and virtualization–you figure out how to deliver it.’

Or they had a recognition that they could pick a package but there were going to be downstream consequences that they were going to have to own up to later and there is none of that.

It’s highly prescribed. It’s the technologies, it’s the vendors, it’s the specification on all of it and it’s like–here, just make this go.

Right, exactly.

Okay so–that’s kind of gloom and doomy. I’m sorry.

I got on my soapbox there. So let’s talk about this for a second. One of the things is you’re talking about containers. Immediately, you’re not allowed to talk about containers without talking about DevOps. As far as I can tell, I think that’s a resolution that passed not too long ago.

Along with the international statute of secrecy for muggles.

Right. And so when you hear people talk about DevOps, they will say all the time, ‘Well, you know with hybrid IT what you need to do is just make sure that you’re using DevOps.’

Sprinkle some DevOps over there.

Go to the store and buy a big box of DevOps-in-a-box and it’s just great. You just turn it on and it’s all magical and life gets better. And the reality is that is not how it works. That is yet another vendor trying to sell you something that is not necessarily going to help if you don’t start with some fundamental changes, and at least embrace things like Agile. I was joking to somebody the other day. I’m going to tell this story, and I don’t mean you have to completely adopt Agile, just experiment with a little bit. So I gave the example the other day, I was talking to an admin on THWACK. I ended up calling him and I asked him a question. I said, so let me just ask you this. Are you in a position where if–let’s say you adopt a sprint–a two-week development cycle. You’ve figured out what’s going to be in it–the requirements that are going to be in it. You’ve got your hard date you’re rolling to. You’re cruising along with your testing. Your implementation’s going to be a piece of cake. And so three days before the due date, an executive comes in and says, ‘Hey, I want to add just a couple of more features to that. We just want to sneak in a couple more changes, and I got permission. So if you don’t mind, we just want to roll that date out about four days.’ What is your answer to that? And the answer,–and I asked this question several times to other people in this situation and they’ll say, ‘Well, well, it kind of depends…’ It’s like nope! That’s automatically the wrong answer. If you can’t say, ‘No, this is our delivery date. Let’s get those new requirements and we’ll roll it into the next sprint.’ You’re just not there. Right? If at the point of being able to have control over the input, your work in progress–not to get all Kanban on you–but some of the fundamental processes aren’t already there–like if you haven’t adopted that culture yet–that is not a fantasy that will magically make this better.

Right and just to identify the mindset. The can I just sneak a couple of things in is actually a symptom of the next rollout’s going to be months away. I need to get these features in because if I don’t, again, back to the SLA, the executive SLA, I need to show I’m transforming the business. And I will not have transformed the business this quarter. Which means I’m not going to see the results of that transformation until next quarter. So I really am going to push with all of my C-level, or my management might. I’m going to push you to take that because I’m afraid. But if you are talking about having another rollout in a week. If you go to that 10 releases a day of DevOps.

Okay, you don’t have to do 10 releases a day.

But I’m saying that if you have a regular release cycle that is more frequent than you release, you have a party, everyone has a drink, you toast, woohoo! Everyone takes a day off the next day because they’re exhausted. Then they come back and they start to think about the next release. And that’s going to be, again, months away–weeks if not months away. If you don’t have that. If you have a structure where the releases are just rolling and we’re always thinking about it–the next one–then there’s less fear factor on the part of executives and management to say, ‘But I have to get my feature in!’ No, no. Your feature will be next week, next two weeks, next whenever.

So remind me in a minute when we get back to that–when we talk about how long we’re all going to have to hold our noses through hybrid IT. There’s a point that I want to make there, but remind me of that point when we get back.

I shall.

Okay. I did say we were going to have just a little bit of demo here. This is just going to be for a second. I want to dovetail this into the–kind of close out the discussion on DevOps by at least mentioning that there are some tools that we will see a lot of times flying in formation with DevOps that are more ‘dev-y’ than ‘ops.’ And for those of you, I don’t know–nobody follows my Twitter, @FerventGeek, but if you did…

EvilMopack does.

Oh. You just gave him a shout out! Now we have to send him a link. And then he can retweet this. But seriously–I had a tweet that I really liked and I really meant it. There was a picture at Cisco Live in DevNet. Right? And it was basically a bunch of guys sitting there doing programming 101 classes. A couple of the classes that they offered last year, two years ago, were a little bit higher. They were things like CLI config and a couple other ones. But this was like Python 101.

Getting started with JavaScript or Python or whatever.

Scripting. What is REST? Basic scripting, basic programming skills–like conceptual programming. And those classes were packed. You guys filled those classes. So when we talk about learning a little bit of dev, we don’t mean all of the concepts of DevOps. But you are going to need to use some of the tools that are available. So let me give you an example when we talk about application performance monitoring, not just server and application performance monitoring, the kinds of things that you’re going to need. So for example, most of the time when we talk about application performance monitoring, we’re really talking about stack monitoring. You’ve seen us talk about AppStack being able to correlate how things are connected. You’ve seen us talk about PerfStack. This view right here. In this case I’m combining things like CPU load, the application performance, along with resources that are out in Azure–I’m sorry this one in AWS–along with storage and everything else. So I’ve got hybrid resources in this view. We typically come at it from the perspective of you have route access to servers, you have route access to pretty much everything that you need, and so you can get all these metrics pretty easily. But when you’re hosting this application off-site, especially if it combines multiple components, isolating issues can be really, really challenging. Right? And people are going to ask for APM and they mean that in the true sense, like code injection, high sampling rate, enormous volume of transactions that can be monitored really, really easily. And so you’re going to start experimenting with tools like APM Tools, so I’m just going to show you this for a second just to kind of give you an idea of where this sort of goes. This is actually Traceview; this is one of the cloud products. This is the hosted cloud native application. The idea here is being able to separate each one of the elements of that application and get inter and inner application performance devs. So if I’m looking at for example, the whole map of this application. This one that actually does, it’s a booking service. There’s a pricing service, there’s an analysis worker, I do authorization, there’s an API and a front end and a booking service. So if I drill in and actually look at one of these services like the booking service, I should be able to see what the performance of the elements are in that service as it’s talking to the other services. How long does it take it to hand off from inside the code of one service to the next? And if I was to look at this for example, I’ve got Spring Framework, my Tomcat and MongoDB and I’m looking at the inner process handoff times on that server.

That’s the thing for those people who are familiar with this usage of the word APM or the letters APM, of application performance management because we use it in a few different ways, it can mean everything from like you said, regular stack, you know, server and application monitoring all the way through. What we’re looking at here is something that is keeping track of the actual transaction as it moves through all the different systems as it goes. How you do that is the magic sauce, we’re not going to get into that. But we’re watching though. We actually start a transaction, a real, live, actual customer transaction that we are watching as it moves through that entire sequence, whether it’s hitting. And the other thing is that it’s not just moving through a single system. It’s not just in Azure and staying in Azure. If it goes from on-premises to an Azure microservice and then back into an on-prem database and then out to an AWS system we want to be able to know how it’s going step by step by step by step, all the way through, that’s the application management we’re talking about.

It is. But in addition to that, because it’s in the cloud, one of the things that a lot of the times that management is asking for is the ability to elastically scale. And so what you’re really looking at for that is this inner process within the elements of a container or a set of horizontally scaled containers. Because you may be dynamically spinning these things up and down and so, you’re not just measuring the performance of the application because with hybrid we’ve moved past monolithic applications to include elements that we burst to. The other thing that they’re going to care about is aggregate performance because this is sample data. So if you have an application that does a hundred thousand, a million transactions or maybe inner process coms per day, you can’t afford to log all of those messages.

No.

So you start to do sampling. So being able to do things like group performance using heat maps to try to figure out where performance issues are get to be really, really important as well. Unlike looking at a regular histogram reporting from a monolithic application where you can go back to the beginning of time. These things are popping up and disappearing, and so what you’re really starting to look for is patterns. You start to use heat maps instead. Heat map for application performance. That’s something that’s a little bit of a different concept and it’s a little bit more like data science. The tools change and not only now do you start looking at things like, hey I’m going to have to actually inject code to instrument parts–to your point of monitoring is the same and it’s also a little bit different–you’re going to have to start instrumenting new things in ways that require a little bit of development. But at the same time, you are still going to have to manage the original stack back in your on-prem environment.

Because it’s still hybrid. You still have that piece of it, but yeah.

So I just showed you an amazing tool, but you’re going to have to learn a little bit of programming to use it and at the same time, you’re not going to be able to transition because these are different skills. These are different skills, these are different types of data and obviously having a single pane of glass to see this would be nice but you’re still going to have to manage the platform and the application stack back on-prem as well because that’s a component part of this

It’s still running. You still have servers and applications local because it is that hybrid IT thing that we’re dealing with.

Right so whether it’s us or anyone else, there will be more tools that are required, and some of those tools are going to require actual, real skills development like learning to program. If you’re going to inject the type of code that’s going to get you the kind of rich, inter-process performance metrics for an application that’s actually running in the cloud, you’re going to have to learn a little bit of Python, or a little bit of Node. Or maybe you have some C Sharp skills that you can carry over into Azure. But you’re still going to have to use those same skills, and that involves debugging and deployment and a bunch of other skills as well.

One of the questions that we get is either, why is SolarWinds investing in the cloud, or is SolarWinds investing in the cloud? We get the question from both different directions. And the answer is that we, of course, are investing in cloud-native tools. Going all the way back, we’ve got now Librato, and Pingdom, and Papertrail, and Traceview. And we have cloud-native tools–cloud-native environment–because that is what people who are in a very cloud-based environment want to see. That’s what they’re comfortable in. Going back to our web dev team–that’s the systems that they’re working in and that’s what they want to see for their monitoring tools.

Except that I would say ‘expanding’ is kind of an interesting word because these products are four and five years old at least already. These are already very mature products. And the other thing is that I don’t even really think of them so much as sort of cloud-native or cloud monitoring as much as they are dev tools. Right? These are the kind of things that–there is a universe of these available–I tend to think that these are four really good ones, but there are a bunch of them out there. And what’s great is they encourage discovery and skills development. These are the kind of things where you can learn what you’re going to need to learn to manage those non-traditional IT resources on the weekends or at night, and they’re kind of fun. Like… …Papertrail for example. Papertrail–I know you guys have heard I do some Raspberry Pi projects–I use Papertrail for all of those things because those things are ephemeral. I blow those images away all the time, and it’s nice to be able to capture all of my syslogs and be able to aggregate those all in one place. But the thing is there’s a free tier. So just playing with these tools on your side projects is a great way to learn them. And I think that’s a big part of when we talk about cloud–is a lot of people say, ‘But I don’t need those skills right now.’ But there will come a time where one: it could serve you well to start using that at least in the small sense in your operation. But the other thing is you may have some new CIO or VP who rolls in and that’s what they know. And then it’s like, ‘Hey, you guys all have six months to come up to speed here.’

Right. So at least playing with these technologies in the background is really, really important. And these technologies being ways to automate the management of hybrid IT, instead of having to do it just through the help desk like you are now.

Right, but there’s a question in all of this which is ‘why.’ Why are we making that investment to go to cloud? It’s not just, ‘It’s the new thing. We like doing the new thing.’ The fact is that we always want to be–SolarWinds is always interested in being where the pain point is, in helping people solve their challenges. But there’s some data behind this.

Ah, yes.

I want to take a look at that. What is it that has been really driving us there? And the fact is that there’s money there, but there’s also a discovery that goes back to what you were saying about hybrid IT and holding our noses–which is there has been, for the last four years now–Cisco has put out their Global Cloud Index. And–so the last one that came out was in 2016. It’s a living document so it’s continuously updated. And what they determined was that by the year 2020–which I have to point out is not the beginning of a dystopian, adolescent novel–by the year 2020, I’ll still be driving the same car I’m driving today, and it actually still will be under manufacturer warranty–so that makes me challenged emotionally. But by the year 2020, what Cisco’s Global Cloud Index determined was that 92 percent of all, all the compute workloads will be performed by a cloud-based architecture. 68 percent of that is going to be public cloud. Why is there such a push to the cloud? Because 68 percent of all the compute workloads are going to be in public cloud. 32 percent will be in private cloud. In addition, overall data center workloads will double from their 2015 levels. So we’re going to have more workloads, which means even more in the cloud than you can even imagine now, or just two short years ago, and the rest of the data that you see here on this screen. So why is there this interest and investment in cloud? Because this is the data that’s coming out.

Okay. So I’m going to put on my skeptical hat here–because I love when vendors and analysts put out these kinds of numbers, and I love the specificity here. 68 and 32 percent–it all adds up. This is just amazing. But, just push this aside and just say this is someone who’s aligning this with a particular message. I think the better data is actually to look at what you guys said, and what a bunch of other IT admins that are in the same roles as you are saying. So we’ve been doing these now for about five years. So it’s not just THWACK, it’s much, much bigger than that. This represents admins that are SolarWinds customers as well as all across IT–IT of different sizes.

Just some quick numbers. The survey this year was over 800 respondents across the spectrum, including over 30 percent that were in the private sector. Sorry, the public sector.

So big government this year, and also international. And so… 95 percent of those surveyed said that they were moving to something that at least looks a little bit like cloud in the next month. But the most amazing thing about this is that 60 percent of them have at least a quarter of their environment in the cloud. So I said before that 50 percent tipping point, right? That this is the first year–but you blazed past that. I think this was in the forties last time. And the other thing that’s sort of a ‘color me surprised’ aspect of this is that we have been watching this change for years, I think. We’ve commented on THWACK before. Every time we release a survey, they ask us to give our opinions on ‘why.’ And we’ve been saying, ‘Hey, hybrid IT or cloud is coming. It’s going to create management problems. SaaS is increasingly deployed–it’s going to cause additional headaches for management and end users. So it at least makes me feel like you are consistent with the answers that you’re giving to these surveys, and that the survey is tracking toward–in fact these numbers are almost spot on–what you were forecasting you would be, in terms of adoption last year.

One point is that 60 percent have a quarter, no more. No more than that. The 40 percent have more than that. One percent are full cloud. Which means of course 99 percent are hybrid IT, of some kind. But you could say, ‘Well, you know if I have 90 percent of my environment then there’s a niggling little bit that’s in the back.’ But what we’re saying is that 60 percent have just one quarter of their architecture moved over to the cloud and the rest of it is still on-premises.

That’s right, there’s a bunch of other data here. We’ll add a link in the show notes to go out and look at it. But there’s also things like, what percentage of you who are running in hybrid environments actually have two or more different public cloud instances; how much of it is public and how much is private; how it breaks down internationally. The data is really, really interesting. So go take a look at it. We won’t talk about it too much. But I just wanted to bring this up because although certainly analysts and vendors are always going to tell you the newest and coolest exciting thing, you are telling us now, for the first time this year, that more than half of you are significantly impacted by this. And so the next question that that leads to…

One more point. And this is interesting. So the thing is that the vendors and a lot of the data–it’s not just that one global survey–is pointing to cloud as being the place to go, which is why you see such an investment. Not just SolarWinds but other vendors are also going to this cloud-native. That’s where all the glory is. That’s where–I started talking about asking the homecoming queen out for a date. The reason why I didn’t get it is because the captain of the football team came in with his dad’s IROC Z, and that technology captured everyone’s imagination. Well, cloud is the technology that’s capturing everyone’s imagination, and it is causing all the investment to go there. But as we’re showing here, the real struggle, and the real point to focus on is this hybrid IT, which is why you see SolarWinds addressing a lot of that in a number of different ways. I just wanted to point out, that’s where we’re going.

That’s perfect. And that actually brings me back to what I asked you to remember to remind me earlier, which is how long–The next question is like, ‘Great. So hybrid IT is the result of this transition to cloud and SaaS and removing previously all co-located resources. How long are we going to have to put up with this? How long are we going to have to hold our nose and wait for this to end?’ And the answer is, well, it is actually going to be forever. And I think one way to think about this is, vendors are not going to stop introducing new applications, new products, whatever else, right? So the difference is, instead of having a lot of heterogeneous applications running your data center, you’ll have a bunch of different stuff in the cloud. So the idea of getting to that single pane of glass–and I know you talk about single pane of glass a lot. That is something that we all talk about. We’re all in the habit of trying to minimize the number of interfaces that we have for monitoring and management. But the reality is that once upon a time, mainframes were a thing. Right?

Indeed they were.

They were a thing, and they were very prescriptive about the way they were used. It was typically one vendor. The hardware was not easy–it was almost immutable. It was a big project to come in and modify the hardware, and so the resources that were made available to it were pretty much static. So the programs were the only things that really changed. And the programming teams were more from academia; they were a little more studious. And so they didn’t have what you’d call DevOps, but they had a much more measured and methodical approach to actually rolling out changes in IT. And also, it just wasn’t that big. Right? And the other thing about it–so it had resources. It was a cool, new thing. Right? And it was especially well funded because people were literally transforming from paper to computers. Like, ‘Oh, computers can do that?’ This was that day. And so the budgets that they had available–just the reduction in cost to the business–what you were being able to do with IT and automation was amazing–and it was a no-brainer to invest a lot there. But once everybody already had computers, and especially once we got to client-server and web and our data centers with lots and lots of products, well, now your competition also has the same gear. So now part of making that work is, now it’s a ‘cost center efficiency play.’ And so, when we look at how we ended up understaffed, and how some solution vendors would tell us, ‘Oh, this one new technology stack for storage is just going to change your life.’ Or all those ROI pitches that they make to management, that somehow this is going to save you a lot of money. You wonder how did we get in this situation where our IT–we’re understaffed and underfunded and, to your point earlier about how long things take to deploy and the complexity, is if we had more hands–if we had time to learn new skills in IT–if we weren’t a cost center where wringing that cash out is–we wouldn’t have those same challenges. So with hybrid IT, you’re sort of seeing this push to cloud now, which just like back in the mainframe days, this is exciting and new.

This is a differentiator. This is a whole different thing than cost center IT.

And it may not even be owned by the same IT organization–maybe somewhere with a fresh, flush new budget. And so they are investing, and it’s cool and it’s exciting, and they’re doing all kinds of amazing things because it’s transforming the business. Sound familiar?

Yeah.

Just like getting computers in the first place? So what happens once we adopt all those new technologies? Business is going to say, ‘Well now everybody has that. That’s commodity, so what do we need to do? We need to roll it up in one department. Where we can manage the costs.’

Pendulum is going…

Right. So the idea that these cool new technologies are somehow going to remain completely alienated and separated is not going to remain. And they will end up being part of what we are doing pretty much forever.

Right. And they’re going to look for the people who have the procedural and technical discipline to be able to manage these things. So yes, we’re going to be living with the technology and this technological reality for a long time, but again, the shiny newness I think is going to wear off and we’re going to be asked to support, maintain, monitor all of this stuff.

I was talking to a guy the other day, one of the new kids–I shouldn’t say that. Should I say that? I don’t know. So cloud everything. You don’t need anything on-premises, you’re just wasting your time worrying about that. Cloud is going to heal all. And I said, ‘Okay. So a user reports a performance issue with the application.’ He’s like, ‘Absolutely. We’ll take a look at our APM stack, we’ll figure out how that thing is performing.’ And I said, ‘Well, what if it turns out to be the access point?’ He said, ‘The access point? Well, that’s part of the delivery network.’ I’m like, ‘Yeah.’ And he’s like, ‘And that has controllers and routers and switches.’ And I said, ‘And there’s a WAN.’ And he’s like, ‘Oh, there’s a WAN and a VPN.’ And I’m like, ‘Yeah, all that yucky on-premises stuff.’

Right.

That’s hybrid IT from the cloud side. There is no way to escape it. You can’t make applications and your users can’t access them.

I had a very similar conversation at one of the DevOpsDays not too long ago, where I was talking about–I was actually talking about living an unplugged life as an IT professional. It was really not a technical talk at all, it was just about how to manage your time and how to be, you know, take some time back for yourself.

I can’t imagine that you would give a talk like that.

I know. Strange, isn’t it? Yeah. So anyway… Somebody came up to me after and said, ‘I just don’t understand about on call.’ I’d talked about managing on call, if you’re unplugged sometimes–like once a week. ‘And I don’t understand why you would need on call at all. I mean if there’s a problem with the system, you just reload it from image and you just reload the container, and you go on. Why do you need to have on call to deal with problem tickets at all?’ And I said, ‘Because I haven’t found anything that will recreate a router from scratch from code yet.’ And he’s like, ‘Oh, network. Right.’ It was just not even on his radar. It was really interesting. So same. That nasty, nasty on-premises stuff–Hybrid IT.

So are we going to survive hybrid IT?

Of course we’re going to survive this. We’re IT. We take it all in stride. We’ve survived much worse than this. [Coughs] Bob. We will get past this. We will learn, we will grow, we will be stronger for it.

They can always count on us to figure it out. That’s what we do.

That’s why we’re IT.

All right. So if we did–just to kind of wrap up. I think we’ve covered a lot. But if we’re going to wrap up, what three things do we recommend are basic skills that they’re going to want to–or basic areas that they’re going to want to focus on?

The first one is you’re not alone. Right?

You are not–you are so not alone. Community, absolutely, is a big part of that. Obviously, THWACK, we see you talking about it. When we look at our survey data, we know that you’re in the throes of this. So reach out to your communities. Remain active and ask a lot of questions, and you’ll find other people in your same domain, your same type of business, who are struggling with the same challenges you are. And it’s nice to know–to your point of you are not a snowflake–that certainly, when it comes to the challenges, increased cost and complexity of hybridizing environments, you are not alone. Everybody is suffering the same challenge.

Right. So group together. Find people who are in your community, again THWACK or others. There’s many good ones. But really start to use that as part of your strength.

Second, the main challenge from a monitoring and management perspective is ‘swivel chair integration’ with new UIs that you have to manage. So get to a single pane of glass as much as you’re able to. And a lot of times, I don’t know if everyone is always up to date [coughs] on their products. But a lot of product capability for managing web and SaaS resources has really improved pretty much across the industry. So if you haven’t downloaded the latest and greatest versions of your products, if you haven’t spent any time with the admin guides and the upgrade list, do that! You’d probably be surprised what’s available in the tools that you already have. And think about using them in new ways.

Regardless of which tools you’re using–obviously, we’re sort of partial to some of the ones that you see on the set here. But the fact is that whichever one it is, you will probably be delighted at where they’re going because everyone is seeing the same data. Everyone understands that this is a trend and you want to see where your vendors are taking you.

And then the last one, of course, would be new skills development. And I think, aren’t we soon to be talking–you guys have been doing some certification?

Yes, I think the next episode is going to be just focusing on certification. Destiny and I are on, what we are tongue in cheek referring to as, Certithon 2017. We’re trying to get all the certs–certify in all the things. But the point isn’t just to collect a bunch of trophies. The point is to start to look at ways to gather skills, and certifications are one way to prove that you have certain skills.

Right. And part of that is it’s one thing to say, ‘Hey, you just need to learn new stuff about new things.’ But one of the nice things about certifications, or at least certain even webinar-based virtual training opportunities, is that’ll help you figure out where to start. Because there’s sort of a natural center of educational gravity around those things that are providing the greatest challenge because it’s the greatest opportunity to help people manage complexity better. So one of the nice things about certifications is these are clearly problems that are big enough that there’s training available and also you’re not starting from scratch. So you can save a lot of time by using an existing program.

Certifications at least give you a list of the questions that you know are questions that you’ll need to answer. Not certification questions. I’m just talking about IT technical questions. Like, ‘Oh, this is a thing. This is something that is worth considering.’

Awesome. [Electronic music]

Well, that was a lot to cover. But I feel like we’ve only just started to scratch the surface. Now, hopefully, this episode helps you start conversations on your team about taking a more measured approach to managing hybrid IT resources than just adding more trouble codes into your ‘help desk drop down’ system.

Yeah, don’t do that. The other thing that we hope is that you gain at least a little feeling of camaraderie with other IT admins. You guys are facing these challenges in earnest now in 2017. At a minimum, we hope that you had a chance to commiserate.

Yeah. Speaking of commiserate, if you didn’t get a chance to get in and commiserate on the chat today, over there, visit our homepage, lab.solarwinds.com and check out the schedule for upcoming events, to be with us live.

That’s true. Also be sure to check out past episodes, especially the deep dives for NetPath and PerfStack–both of which that you’re telling us to take some of the sting out of hybrid IT.

‘Sting out of hybrid,’ that is my new band name.

I thought it was ‘Linux Desktop Grokker.’

Oh yeah, like yours is any better?

‘Heptagonal Glee Club?’

That is a Kubernetes joke, isn’t it?

Oh, yes. Yes, it is.

And with that, I’m Leon Adato.

And I’m Patrick Hubbard, and thanks again for watching SolarWinds Lab. [upbeat electronic music]