Virtualization is one of the well worn tools in IT Management. To find out what one vendor has been seeing in the virtualization landscape, I talk with Convirture’s Arsalan Farooq. We of course talk about the company and its products, but we spend most of the time talking about the trends in usage and management practices Arsalan has been seeing.
- History of ConVirt – and check out the two part screencast from last year that I did with them.
- I ask Arsalan what admins are actually doing when you manage a virtual environment? What types of activities entail “virtualization management”?
- What’s the decision process for going with open source virtualization vs. closed source? Arsalan says that VMware does well in the Windows world, but not as good at Linux capture. Silos of Windows and Linux.
- Use of private cloud in enterprise… vs. MSP’s – as I note, I’ve seen a huge rise in service provider interest for cloud and IT management technologies.
- We discuss “cloud” vs. “virtualization”? What’s the difference between the two, the overlap, and so forth.
Michael Coté: Well, hello everybody! It’s the 22nd of September, 2010, and this is a special edition of the IT Management & Cloud Podcast. Instead of the usual co-host – you may remember John Willis on here – instead of us going over the news and events that’s going on at the moment, I have a guest on and we are going to talk about the wacky fun world of virtualization, if you will.
As always, I am one of your co-host, Michael Coté, available at peopleoverprocess.com. Why don’t you introduce yourself guest?
Arsalan Farooq: Hi! So this is the guest, otherwise known as Arsalan Farooq. Glad to be on the show. I am the CEO at Convirture Corporation. We do virtualization management for open source platforms. So Xen, KVM, and the upcoming cloud platforms. Glad to be here, big fan of your show, and looking forward to talking some shop.
Michael Coté: You gave the sort of one line intro there in your introduction to Convirture, but can you give us a sense of like what exactly ConVirt is, I mean just give us the kind of, I guess I would call it the medium deep dive.
Arsalan Farooq: Okay. So medium deep dive. So basically let me start with a little bit of history. We started off in early 2007, and we kicked things off with an open source project, which was at that time called XenMan, which later on we renamed to ConVirt.
What that was basically was, essentially we were targeting coming up with what we at that time thought was enterprise class management for the open source platforms, and we came out with XenMan and then ConVirt in the open source arena. Baked that, took that through two generations, until earlier this year we launched the Version 2.0 of ConVirt.
And what ConVirt is basically, I mean, I think the simplest way to put it is, for those of you familiar with virtualization, if you think of VMware’s VirtualCenter Suite, the VirtualCenter management toolset, we basically do that for non-VMware platforms, specifically open source platforms. So that’s what we were targeting with ConVirt, and that’s what we brought out in the open source arena, against open source hypervisors earlier this year.
Now, move the clock forward, around six or so months, as of late July, we also introduced our first commercial offering, which is the ConVirt 2.0 Enterprise edition. And this has some interesting new features, which are more targeted towards large enterprise-wide cloud and virtualization deployments.
So basically to cut it short, we have been doing virtualization management in open source for many years, and very recently we have now brought out our enterprise offering, which we can talk about a little bit more hopefully in a little bit.
Michael Coté: Yeah. And for people who are like mega interested, we actually did a couple of screencasts a while ago, it was probably like a little under a year ago or so, that kind of went over what was going on at the moment. So there is plenty of demo stuff to see out there.
This is kind of like — it’s always fun to ask sort of basic questions. But what are the — so if someone is using your stuff kind of like daily or weekly or whatever, what is the management that they are doing of virtual stuff? I mean, what are you actually doing when you are managing a virtual environment?
Arsalan Farooq: Yeah, that’s a great question. So think about it this way. So you are sitting there and you have decided that you have got a bunch of Linux in your environment and you want to virtualize your servers. So the first part is easy. Where do I get the virtualization platform from? You get it from Linux, you get it as part of your Linux install with KVM or Xen or what have you.
The next part is, okay, well, great, I have deployed it. That looks very nice. What do I do next? What’s next? Where do I get virtual machines from that I am going to run on this environment? Virtual machines are these virtualized operating systems that you run on your virtualization platform. How do I create virtual machines? How do I provision them into the environment? How do I start them, stop them, basic administration tasks? How do I do migration, which is a high value feature that virtualization brings.
Things like template-based provisioning, schedule provisioning, monitoring, lifecycle management, all of these things, that once you have the platform down, your day-to-day tasks of provisioning, deep provisioning, administration, that’s what the basics ConVirt’s management toolset brings.
The way I think about it is, you have got the platform and then you have got the day-to-day operation and management of that platform. The platforms are easy to find. Management is somewhat harder to find, especially in the open source area, and that’s what we brought initially with the ConVirt Open Source package I should say.
Michael Coté: I mean, coincidentally, this morning I was talking with someone about some open source survey data that they were doing about management software [Zenoss’ recent survey], and of course virtualization came up. I think the top three hypervisors that they found among their survey data was — I mean, VMware was a big leader, which is clearly not open source, but then after that I think was KVM and Xen, each of them holding about 20% among their survey base of users [actually, I believe it was 20% total, for both, not each].
So I mean, there is a — you throw together KVM and Xen and you actually have a pretty, at least among their base, a pretty significant among of virtualization using that. I wonder, you guys are basically servicing the kind of open source crowd, the first question I am kind of curious is on this kind of area is like, what do you think — what motivates people to go with an open source sort of virtualization solution versus a proprietary one? I guess by proprietary I mostly would mean VMware and Hyper-V and things like that.
Arsalan Farooq: Yes. First of all, the data you are describing, we have some internal data that we have developed as well as some of the analysts that we talked to have been doing so. The numbers you just mentioned around Xen and KVM are remarkably consistent across most of the studies we have seen and our own experience more importantly. So I couldn’t agree more.
I mean, the effect is certainly there, and really that was the market opportunity that we wanted to address. The gap in the market where VMware has a fantastic management vertical and when it comes to tremendous new platforms like KVM and Xen, there was nothing, and that’s where Convirture comes in and ConVirt comes in.
Now, the second part of your question, which is what — we run into a lot of people, and you characterize them as open source crowd, but actually we are seeing, not just the traditional open source aficionados, but also large Fortune 500 enterprises that have existing VMware deployments, those two are now starting to look at Xen and KVM.
Now, when we talk to them, there are myriad reasons why people are looking at open source. Well, the first of them is really the way the large data centers are structured. So if you go in and forget about virtualization, just walk into a large Fortune 1000 data center and what you would find is essentially a Windows shop and a Linux shop.
There is this entire Windows part of the data center that they run their Exchange servers and their IT infrastructure, they run a bunch of .NET applications that they have developed and all of that stuff.
Then you have got the Linux part of the data center, where guys are running Oracle on Red Hat or they are running JBoss or a bunch of other stuff is happening.
So I tend to think of them as the Windows silo and the Linux silo inside the large data centers. Now, VMware has been extraordinarily good with capturing the Windows part of that data center when it comes to server consolidation and virtualization.
VMware’s presence for the Linux side does not have the same level of traction. Now, there are lots of reasons for that and we can go into that, Coté, if you want, in a bit, but the key point is that you have got the Linux silo, and the guys that are running that, the admins and the IT administrators and the decision makers, what they have is they have their favorite distribution or their favorite set of distributions, be that CentOS or Debian or something, or SUSE, Red Hat, or what have you.
And they have a lot of administrative processes, they have a lot of expertise, they have a lot of provisioning and other infrastructure in place, and what they see is, hold on, once we start looking at virtualization or make plans towards virtualizing our part of the data center, we don’t have to move to a new platform, we already have Linux, and Linux has a virtualization platform, be it KVM or Xen, depending on which distribution they have, built in. So there is no change. They can just activate that and go with that.
So we hear a lot of — we have a lot of conversations where people come to us and say, look, we are now ready to virtualize the Linux side of the data center. We already have expertise in say SLES, SUSE Enterprise Linux, and we love the prospect of actually not having to change anything, just turn our virtualization on inside Linux. And our problem right now is that we don’t have anything that looks like virtual center that we have on the Windows side.
And that’s where we say look, ConVirt and ConVirt Enterprise will sort that problem out for you. So that’s one of the key structural reasons, I think, what we are seeing.
Michael Coté: No, I think that — like the, to use the word, the “silo” sort of basis there, it’s actually kind of fascinating in the sense that, I think it still seems to be in that phase — virtualization still seems to be in the phase where there is a bit of a toss up of sort of who is in control of it. I mean, there’s sort virtual admins and other stuff, but it’s not quite as ironed out as like DBAs versus network versus security, so to speak, which are very fortified organizational areas.
No, it does make sense that if you divided up Windows and Linux and UNIX and Mainframe, I mean, the bigger the company, the more divisions there are, and that you would sort of — it would be hard to kind of sell across all of those, and to kind of create an opportunity for the Xen and KVM based stuff to come in, like yourselves, so that’s kind of curious.
Arsalan Farooq: Yes. And let me just add a couple of points to that. I mean, I think that one thing that all the data that we have, as well as have developed internally and otherwise heard, is that, there is tremendous amount of heterogeneity when you look at the Fortune 1000 set or really anywhere else. This perception of a complete singular, homogenous stack inside an entire data center is just — it’s a pipe dream, it has been a pipe dream. I have been doing systems management for the better part of 15 years, we have never seen it. There is always heterogeneity, and in the case of virtualization, these are some of the drivers.
Let me just add a couple of other things to your previous question, because I think it’s worth spending a little time on. In terms of what other drivers are we hearing from customers who are evaluating Xen and KVM and perhaps go a little way towards explaining why we are seeing this Xen and KVM data that you were referring to earlier. So that’s one thing, the structural elements and the way the organizational structures are and the way they are planning going forward.
The next thing that I hear, and this is not really talking to the admins, but rather talking to the CIOs and the decision makers, what you find is that there are some tremendous cost drivers that are built into, that move into, looking at community or commercial Linux. The way I think about it is, the same drivers that drove Linux into the data center are the drivers that are driving Linux virtualization into the same data centers.
I have my own theory of what those drivers were, but when you talk to customers, they say, look, we are looking at this, we want the server consolidation, we also are interested in the enterprise cloud potentially going forward. It’s hard to size what kind of environments we have, especially for those CIOs that are thinking in terms of the enterprise cloud or the “private cloud”. And they want to make an assessment based not just on technology, but also the cost scaling that their environments will have to have.
And we find that the way some of the proprietary platforms, like VMware and others, are priced and their business model sort of gives the Linux based platforms a tremendous cost scaling advantage.
So this is something, once again, that we hear and it makes sense, and yet another one of those drivers.
And the final stuff that I would say is, we have had some conversation where VMware has done a great job, all credit to them. For the past few years they have had a hammerlock on this market. I know it’s fashionable to say this, but I am going to just drive this from what we are hearing from customers, there is a real concern about vendor lock-in out there.
Michael Coté: Right, right, right. Following up on those, like — when I talk with people, both on kind of like the sell and the buy side, if you will, with vendors and people who are using technologies, it’s interesting to see the kind of rankings of — like you are talking about private cloud and license issues, and it’s interesting that license management is, year over year, is always like a top concern of any IT management thing.
I mean, on the one hand, there is like the annual fear of getting audited and owing money. So you always want to make sure that you are doing the right thing essentially, you are pirating software or whatever unintentionally. But especially in the — it would interesting to hear how real or how much private cloud stuff you see out there, but when it gets to the more — further along the cycle of maturity of using virtualized infrastructures, or like you were saying, there is a lot of migrating going on and scaling up and scaling down, and things that you would kind of label as cloud, if you will, at that point, license management sort of becomes a big hurdle to get over.
And short of having an enterprise license agreement with Microsoft or whoever else it is, they are usually the source of license issues, or it could be Red Hat or whoever else that you are licensing stuff from. It seems open source kind of like takes that issue away if — as long as the sort of any commercial arrangements you have are entangled with that.
So I mean, it is interesting to see that dynamicness of virtualization kind of driving companies to rationalize their licensing stuff.
Arsalan Farooq: Indeed, indeed. And I think that that is most certainly the case, the license tracking, license management, as well as — and I am going to — this is something that we have been seeing recently, especially since we released Enterprise, which does have a licensing model associated with it. This is something we ran into, where there are two aspects to the licensing; one is the one you are referring to, which is, am I compliant? Am I violating something horrible?
The other thing is that the licensing model that you are faced with from the existing proprietary vendors and the cost elasticity of that; and what I mean by that is, if I were to go into a “private cloud” type project, by the very definition when it comes to that, as you were referring to the clouds, they are elastic entities. And to date, a lot of the pricing models that have been available, licensed pricing models from the proprietary vendors, have not been conducive to doing any type of C-level planning when it comes to cost scaling in a cloud type deployment.
When we were designing our licensing scheme, this was something that we were acutely — we went out, we talked to more people than I remember. And one signal was very clear, look, if you guys are going to enter this cloud management business, make sure that your licensing model reflects very, very finely what an actual cloud deployment is looking at, in terms of elasticity scaling up and down.
Michael Coté: Oh, right, right. And for example, if you had an agent on each machine, if you yourself started deploying something, you want to make sure it matches that same topology.
Arsalan Farooq: Absolutely! And the tracking aspect is there, but let me give you, perhaps, if you are interested, I can give you a couple of other interesting examples.
Michael Coté: Yeah, yeah.
Arsalan Farooq: So think about it this way. One thing that we ran into was, you asked me, what are we seeing in terms of the private cloud or just cloud in general, so let me start with that first.
So what we are seeing, and this is a pretty consistent pattern that we are seeing, is that, when it comes to the Fortune 1000 set, the private cloud or enterprise cloud, as long as we know what that means, because the definition themselves are a bit up in the air at the moment. But what the CIOs are asking is, okay, well, this looks interesting, I have heard a lot about this, a lot of vendors are of course positioning themselves, this and that. There is a considerable amount of confusion in terms of what it is.
But what we are finding is that most CIOs do know that they — they sense, they smell the fact that there are certain parts of their operational models that could be amenable to go to a cloud type operational model.
And the next question that they ask is, okay, well, that’s good, but how does my existing — what does that mean for my existing infrastructure? What does that mean in terms of migration and all of that stuff? So they are actually asking these questions, and that’s the stage we find them at.
I have yet to have a conversation with a CIO who is saying, look, I know exactly what a private cloud is, I know exactly what that will give me, and I am, guns blazing, going right into that.
Michael Coté: Right, right. They are sort of in the planning — the planning evaluation research phase, early planning phase, if you will.
Arsalan Farooq: Early planning. So that’s what I am finding. That’s what we look at when we talk to the CIOs. But there is a very interesting market where the awareness of the cloud and the cloud operational model is much more advanced than that. And that is the hosting market.
Michael Coté: Yeah, yeah, with service providers and hosters and all of that.
Arsalan Farooq: The hosters, right. These guys, look, one could argue that they have been doing private cloud or hosted private cloud for ages, right? I mean, Convirture is, we are a four year old company, but still all our IT infrastructure is, is outsource into an ISP hosting provider.
These guys had been in the business of providing VPSs, which is Virtual Private Servers, which is essentially selling not just applications for rent, but also compute for rent, for a while. These guys are in my minds more on the forefront in their thinking, in terms of what having a cloud management tool from our perspective means.
When we talk to these guys, you get a good signal on what kind of licensing models, bringing it back to where we started, could work.
So what we did when we went out and built our licensing models was that, we basically went and talked to this hosting provider segment, and talked to the enterprise segment, and what we found was that, when you talk about the cloud inside the Fortune 1000 set, versus the cloud in terms of the hosting set, the requirements are very different. The hosting set actually came out with stuff, because they are used to doing this stuff, VPS and providing cloud compute, as well as platform functionality, that they had some very interesting input on how a cloud licensing model would work.
So we actually found ourselves in a position where we ended up developing two different licensing schemes; one we call enterprise volume licensing, and the other one is called hosting provider licensing, to address the two needs.
The reason for that is that, the procurement guys inside a large IT environment, like in a Fortune 1000 company, think along different lines. Their decision variables are different. They think in terms of hardware procurement. They think in terms of buying capacity and buying overcapacity, and those kind of things. So they like to make decisions upfront in terms of what they will buy, how much they will buy, and the purchase cycles are longer.
When you go to the hosting market, the situation changes, these guys are high-volume, low-margin businesses. They do not have a priori knowledge or visibility into what their compute requirement or what their capacity requirement is going to be. They want their spend on infrastructure to closely mirror what actual capacity they have in play.
What we did was, we said, okay, well, how would you like it? So they said, well, we don’t want per socket pricing, we don’t want per server pricing, none of that works for us, because we don’t know what number of sockets, what number of servers we are going to have in three weeks, much less a year. So give us something that we sell, make things match what we sell.
So we came up with a per virtual machine pricing model for them, because that’s something they sell, and that’s what they can compute in terms of margin.
Michael Coté: Right, right, which gets back to your point of like matching the licensing to the usage of it.
Arsalan Farooq: Absolutely! Because otherwise, I think that the big proprietary vendors, we know who they are; VMware and to a lesser extent Citrix and Microsoft, they have basically been — their pricing models, their licensing schemes, have been somewhat rigid when it comes to the cloud models that are emerging out there. I am talking about both the cloud operational model as well as cloud cost assessment and capacity planning model.
I think that one advantage we had was that, we were able to learn from the experience that we were observing in terms of the proprietary vendors dealing with these kind of guys. Bringing this back, I think that the hosting, the signal that we are getting from the hosting market is actually a good indicator of what is going to happen once those CIOs we just talked about, who are still in the comprehension and planning phases, once they start deciding how they would like their enterprise cloud deployments to be built out, and what kind of vendor topology and licensing topology they would like to deal with. So we are treating the hosting as an advanced signal.
Michael Coté: Yeah, that kind of mirrors what — ever since about a year or so ago, amongst the — putting it very broadly, amongst the IT management people, like companies and offerings that I looked at, spanning like monitoring and virtualization and even cloud stuff, and to a limited degree kind of like, I don’t know, service management, the process side of things, there’s been a pretty noticeable like, I don’t know, I guess you would call it subculture of selling to service providers.
There is definitely — yeah, I mean, I haven’t really like sat down and done a thorough study of it, but I feel like it’s kind of like what you are getting at, is — the angle I usually have that I talk to people about is, if you think about being a service provider, like it’s kind of an undifferentiated commodified thing that you are offering. I mean, you basically — the prices always have to be very low, like you were saying, high volume and thin margins, and you can compete on quality of service and customer service and things like that.
But it’s only recently that folks like yourself and others have come along with things that are affordable enough and sort of priced accordingly that you can actually start to pull in other features that you provide.
And if you look at, like Rackspace has their OpenStack thing out, and there’s other cloud offerings, and then there’s various monitoring tools, and tools like yourself, and you can kind of see that those start to become features that service providers can sell. I mean, basically it’s an up-sell opportunity, if you will, and it’s just a nice to thing to have to sell.
Arsalan Farooq: Yes, absolutely! I think that as far as Convirture is concerned, as far as our view is concerned, we are not just a hosting provider business; we are basically a hosting provider targeted business, nor are we just a Fortune 1000 business. To us, we are looking, when it comes to cloud, I mean our strength of course remains in traditional virtualization, Xen, KVM, the way we talked about earlier. But as we move to the cloud and as we provide more and more cloud centric features, the first set of which we released with Enterprise 2.0, we are watching very carefully what is actually emerging.
I mean, Coté, you know how much noise there is about the cloud. I mean, forget about what it is and what its benefits are, I mean the definition of the cloud itself is right now up in the air.
Michael Coté: Right.
Arsalan Farooq: But what we are trying to do is, out of all that noise, we are trying to pick up the signal, and we are trying to see, what are we hearing from the 1000 set, what are we hearing from the hosting set, how much of that is domain specific, how much of that is domain nonspecific? And make sure that as these platforms, OpenStack, other platforms that are out there, even proprietary cloud platforms that are out there, what are the actual, the real management needs that actually are required to have anybody; be it a hoster or be it a Fortune 1000 set, transition their traditional IT operations to a cloud operations model? What is required from the management tools? And that’s what we are trying to track.
Michael Coté: Along those lines, like it will be interesting to see kind of — since you have a foot in both of those worlds, virtualization and cloud, like I sort of see — I don’t know, it seems like those are two like ends of the same elephant, to use that metaphor, if you will, or at least they — I don’t know, maybe it’s some mythical beast that used to be two separate animals that’s now joined together, and it’s two separate ends of that. But it is — I mean, how do you guys rationalize out, this is virtualization management and this over here is cloud management? Like it seems like an interesting thing to kind of sort out in your head.
Arsalan Farooq: That’s a great point! I mean, I think that we don’t, we agree. I mean, I am not sure what the mythical animal is, but I do know that they lie on a continuum.
Michael Coté: Right.
Arsalan Farooq: To us, this is not a step function, nor is it two different areas. To us, it is a continuum, and I will tell you why I say that. When you look at what cloud is, I don’t like to go into and start picking arguments with people who feel cloud is this, cloud is that, but I will just state what I think it is.
I think cloud is not a technology. Cloud is an operational model. You can have a bunch of technologies that make operating in a cloud operations model easier or harder or less expensive or more expensive. But I think it’s a mistake to say that cloud, especially the enterprise cloud or the private cloud, is a technology.
The technology already exists to enable — the core enabling technology to enable enterprise cloud computing already exists and has existed, it’s called virtualization. That’s the technology aspect of it.
The difference between virtualization to do server consolidation versus virtualization to do rapid, dynamic, provisioning, de-provisioning, and elasticity, which is a cloud operations model, with potential self-service and all of that stuff, these are operational models that need to be enabled on the technological core, which is virtualization.
So when it comes to how we see it, we see this as a continuum. We see this as, look, virtualization came out, it’s an enabling technology. For the next five to ten years, we will still be finding new uses of how we can extract value from virtualization.
In the first wave, the lowest hanging fruit was server consolidation, energy savings, power savings, all of that good stuff that’s been happening, and that’s been driving a lot of the adoption. Now that that’s well understood and that’s what we are managing right now for the open source, the virtual center for the rest of us basically.
Now what’s happening is, the next wave of realization is emerging, which is that, look, there is another thing that we can do with virtualization; it’s called the cloud operations model.
Now, what is that? Well, we can talk a lot about that, but what it isn’t is, oh, we need a new technological substrate that will come in and we need to have a new platform, virtualization is old, we are now talking about a new platform, that will be the cloud platform.
This I don’t agree with, it’s patently incorrect in my view. What we are talking about right now is, we have done, we have understood, we have extracted value out of server consolidation, now we are moving to the next frontier in terms of extracting — on our journey to extract value out of virtualization, the technology, and that is this elasticity, this dynamism that — and this highly efficient provision, de-provision, on-demand, compute, as well as topology that we can do in our data center. And that’s where the next set of operation models are coming from.
Now, to me, this is just a continuum. A management tool that is managing your existing virtualization deployment, which you happen to be using just for server consolidation, and now want to use for something else, should be the same management suite.
I mean, you shouldn’t have to rip out your management and your platforms and change your administrative and business processes. You should say look, today I have virtualization, I have virtualization management. Today my operational model is server consolidation, tomorrow my operational model is to add, not replace, add cloud operations as well. And that’s how we see it on a continuum. Hopefully that made sense.
Michael Coté: Yeah. No, I mean, I think basically what you have got there is a maturity cycle or continuum or whatever, and it’s sort of — and it kind of matches what I sort of see in practice than in theory, the sense that, you use virtualization basically as a cost savings thing for consolidation, and stuff is cheaper without spending more money, which is a crazy thing for IT.
And then once you kind of get to that, the next like load of gold you mine out of the river or whatever, is basically, it’s kind of an intangible kind of thing, but it’s really like the flexibility and it’s easier to manage, and because everything is virtual, it’s not quite as onerous and you can start to do things a little faster and so forth and so on. So it’s kind of like, we do our job better, not just cheaper.
I kind of feel like that’s where most people are at the moment, and kind of like the sort of visionary, if you will, or the kind of out there thing is kind of like, so how would we actually use this infrastructure to like deliver applications much faster or like to actually do something that helps generate revenue, not just save revenue or make sure we don’t spend too much revenue, and that’s kind of where things start to get exciting for me.
Arsalan Farooq: I completely agree. I mean, I think you have couched it very well here, that so far virtualization’s value; remember, technology remains virtualization, there’s no new technology here. The basic revolution is virtualization. Everything else is just accreted.
Now, having said that, your point is very well made, that so far virtualization has been, people have been extracting essentially bottom-line gains, but now they are realizing that there are some top-line gains here. And that’s what we are beginning to see in the public cloud and in all these enterprise cloud models, and stuff like this.
I am going to come back to my message. To anyone who is listening to this and is confused about virtualization, I will give you my view. My view is, you have the technology; there is no need to build a new platform, it’s already there, what you need is the cloud operations model, and is that valid for the type of stuff I do?
And this actually leads into sort of something that dovetails with our view on what — this continuum that we were talking about, virtualization being the base and the various loads or veins that you are going to mine as we go forward.
One of the key things we have to think about here is that, when it comes to cloud, and I am talking about enterprise cloud or private cloud, what have you, I don’t know — whatever your preferred term is. On-premise cloud, let’s call it the on-prem cloud, in a Fortune 1000 set. The on-prem cloud is not a panacea, it’s not an answer for everything you do, it is relevant for a subset of things that you do.
In fact, there are things that you have in your environment right now, may be that could benefit from server consolidation, but will not benefit from transitioning to the cloud. Let me give you some examples.
You have static infrastructure inside your dataset, your DNS servers; maybe even your Exchange server, maybe things that are static and singletons in your environment, your routing infrastructure, that stuff is not elastic, it is not provisioned or de-provisioned on demand. It has absolutely no sort of either compute or topology dynamism inherently built into it. This stuff can benefit from server consolidation because the power gains and all of that stuff, but it doesn’t necessarily have to go and be provisioned in some sort of a cloud style deployment, because you do it once and it’s over.
Now, there are other things you do in a Fortune 1000 set which actually are amenable to a cloud operations model. Your QA environment comes in and provisions and de-provisions these environments to run QA cycles, your job processing cycles, which have a timetable or are seasonal, that have seasonal variations in them.
There is compute that is dynamic and has a seasonality or an ad hoc nature to it, and then there is static infrastructure that gets deployed once and never gets used.
So we argue that, look, the dynamic workloads that you have can benefit in many ways; and we can go into those if there is interest, from a cloud or an on-prem cloud model, the static stuff doesn’t need to.
So what we see is that, going forward, the enterprise cloud or the on-prem cloud, on-premises cloud, is going to see a tremendous penetration, but it is going to see penetration driven by a certain type of workload.
I have seen things been said and written out here which says, oh, the on-prem cloud or the cloud model is going to go, throw all your IT practices out, start a new rip and replace. And I look at that and I say, not only is it unrealistic; in other words, if you walk into a CIO’s office and say, I have got something, rip and replace, and you are going to have Nirvana, he is going to throw you right out with a security escort, right?
But not only is it unrealistic, but it’s actually unwarranted. It’s poorly motivated. It doesn’t reflect how at least we see things are going to go.
Michael Coté: Yeah. I think to that point: I think there is a certain — if you are going into a sizable enough IT shop, there is going to be a lot of legacy stuff, and there is going to be a lot of things that you need to bring along on the journey, so you can’t just sort of like start from scratch, if you will. So most rip and replace things don’t really take into account — at some point in time the thing that you are ripping out worked pretty well and did its job and you have got to make sure that you keep doing that. You can’t just do things only the new way essentially.
Arsalan Farooq: Right. That I think is just a pipe dream that new vendors and a lot of commentators share, but it just doesn’t happen. So I want to bring that back to where Convirture stands on this. We are firmly in the virtualization management space, we are firmly in the cloud frontier; at least that’s how we see ourselves and hope our customers see us.
But I want to emphasize that we are trying to take a much more realistic approach to this. I mean, our entire executive team in our company, guys that have been doing systems management since the early 1990s, so we have been around long enough to know that the challenge here is not to come up with stupendous sounding models and then trying to convince the Fortune 1000 set that this is the thing for you. How many, Coté, times have we seen this before, right?
Michael Coté: Right. I call it the “motherhood and apple pie” sort of messaging, like, oh, yeah, it’s everything you ever wanted, except it’s going to work this time.
Arsalan Farooq: It’s going to work this time, and, if you just rip out all the stuff that you have which doesn’t work and put this in, everything is going to be fine. No.
And I think the problem is a little more difficult than that. I mean, I personally don’t buy into cloud for cloud’s sake. I buy into — the question I ask is, okay, we understand what the cloud operational model is, what are the benefits that I can walk into a CIO’s office and tell him he is going to get realistically, and that’s italicized, in terms of, I am not asking him to do something that is not — something that he or she can do. I am saying, look, I acknowledged that you have an environment today. I acknowledge that you are interested in seeing what cloud can bring to you. Not only can I tell you what cloud can bring to you, but I can also tell you how much of your problem set will it cover, the benefits, and I also owe you the response to, well, how do I get from here to there. That’s really the management philosophy that we are trying to proffer with our Enterprise edition.
Michael Coté: I mean, it’s sort of a nice pragmatic approach to things. You don’t want to stick with what’s old necessarily, and you don’t want to like boil the ocean and like rip up all the turf and plan in new stuff essentially, it’s a nice middle road to go.
Arsalan Farooq: Right, and driven by not our religion, but their need.
Michael Coté: Right, right, definitely. Well, I think that’s a good place to wrap it up. I mean, I was looking forward to — from several times we have talked before, it’s easy for you and I to get derailed in the sort of official briefings we have just talking about the industry, if you will. So thanks for giving me the opportunity to capture kind of that, just chatting about stuff. I think it was good stuff.
Arsalan Farooq: Well, thank you so much for having me Coté. As you have mentioned, we always have great conversations, so hopefully we will have many more to come.
Michael Coté: Definitely! We will see everyone next time.
Arsalan Farooq: Alright, very good! Thank you. Bye, bye.
Disclosure: this episode was sponsored by Convirture, several companies mentioned above and in the area of IT Management and cloud are clients, see the RedMonk client list if you’re interested.