Finding Sunny Skies in a Cloudy Environment
A Conversation With Enterprise Management Associates’ Dennis Drogseth
Imagine trying to hold onto a cloud. As soon as you wrap your arms around it, it changes shape and slips away. The best you can do is blow it in the direction you want it to go, and hope it stays intact. It’s not quite that difficult – but not that far off, either – to manage the digital cloud that is taking over the enterprise landscape. The cloud promises cost savings and agility that businesses are eager to capture. But it also is pushing IT departments to evolve their perspective and operations to accommodate the new paradigms, and to engage in dialog with more areas of the enterprise than ever before. Enterprise Management Associates Vice President Dennis Drogseth is closely following the track of the high-pressure system that is the cloud, and recently authored the EMA research report, "Operationalizing Cloud: The Move Towards a Cross-Domain Service Management Strategy". Benchmark sat down with Drogseth to learn about the impact of the cloud on enterprises, how they’re managing it, and the critical importance of cloud performance, both financially and operationally.
Benchmark: Let’s start with your 30,000-foot view of the cloud today – adoption in the enterprise, trends – what’s the big picture?
Dennis Drogseth: The big picture is that most companies have figured out– especially when it comes to end-of-life hardware – that it’s time to move to a virtualized infrastructure. That can bring a lot of value. It’s really more of an internal cloud direction, and aside from the CAPEX [capital expenditure] savings in terms of buying all that hardware, there are usually a lot of other improvements.
If they’re well-organized and well-deployed, these internal cloud initiatives benefit in terms of service resilience because of failover and the ability to switch around on different VMs based on usage requirements. You don’t have to bring a device down for an update, for a patch or a reconfiguration. You can do it much more dynamically, so consistently, then, the service stays up. And that’s been a big plus.
A lot of the infrastructure, data center virtualization efforts have moved past the 50 percent mark to 60, 70, 80, even 90 percent. Very few are 100 percent and some may never get there for various reasons, but the argument now has to be, ’Show me why we can’t virtualize the software-hardware infrastructure to support this application service.’
Probably the leaders are at 70 or 80 percent virtualization. There are quite a few that are more at 10, 15, 20 percent. There was a little bit of a wall between 25 and 30 percent, and that was, I think, in part because a lot of the initial efforts at virtualization and cloud were not necessarily done by the proper strategic owners but by people who wanted to experiment.
Benchmark: You recently completed some important research on cloud operations. What did you learn?
Dennis Drogseth: In December of last year, 70 percent of our 155 respondents said they had to redo or rethink their cloud initiatives, and part of that was moving towards a more strategic ownership.
We’re seeing the dramatic rise of a new group – ’cloud and virtualization services’ – which is trying to get its arms around all the various pieces of cloud and work with the executives across domains to get everything to work together. That’s kind of new. I would say a year ago, that was not the case. A year ago we were looking at a much more piecemeal, less executive, more siloed IT-centric approach.
By the way, I probably should add that Enterprise Management Associates has spent no time really trying to define cloud. We use the NIST [National Institute of Standards & Technology] definition, partly because a lot of other analyst firms have spent a lot of energy to come up with their own spin, and partly because frankly, in my experience with IT, they don’t care. Who cares what you really call it? Is it truly cloud or is it not?
So having said this, software as a service always leads infrastructure as a service and platform as a service. Our data indicates that about 77 percent of those who feel they have cloud initiatives have software as a service deployed or about to be deployed. Whereas for infrastructure as a service and platform as a service, it’s about 39 percent.
Now I would argue that, in reality, if you were strict about ’is it true cloud or not,’ probably 70 of that 77 percent would get blown away because true NIST definitions for software as a service would include a purely on-demand usage that requires on-demand billing, which in my experience almost nobody’s doing.
Benchmark: OK, so that would be a pretty strict interpretation.
Dennis Drogseth: Yes, and again, I would say that’s interesting to note, but it doesn’t matter if it doesn’t fit someone’s pure platonic definition of cloud.
Benchmark: Right. So what is important in defining the cloud?
Dennis Drogseth: What’s interesting about the cloud – aside from the fact that it’s in many respects an ugly hodge-podge of different technologies that people have thrown together under one word – is that in spite of that, what is consistent about it and valuable about it is it’s a more dynamic system. It’s more cross-domain. You can’t look at things just in single silos.
And it’s forcing IT to take on a different role, which is more of what I would call a broker of services than a territorial landscape where you have academic skill groups – the network group and the systems group and the database group – like departments in a university owning territories and then trying to get those groups to work together.
Cloud is accelerating the trend to move to a much more top-down, service-centric view, to break down the walls between some of those skill sets – not to do away with the skill sets but to create process flows and dialogue – so that they actually can function less as independent tribes and more as people who know how to work together.
Benchmark: For the good of the enterprise.
Dennis Drogseth: And that is a great thing. Ironically, in my opinion cloud is a catalyst for this, because I don’t believe cloud is an end goal. Cloud is a set of enabling technologies.
Benchmark: To paraphrase, many organizations started off in a kind of an experimental mode, but have realized that it is operationally and strategically advantageous to move more and more to the cloud. So one of the challenges has been how to, in a unified fashion, strategically integrate it into their operations and manage it in total rather than in piecemeal.
Dennis Drogseth: Right, and that typically means a hybrid approach, which can be cloud and non-cloud, public/private, infrastructure plus software as a service hybrid. The term ’hybrid’ actually has at least five flavors. I like to use the term hybrid cloud environment, and then I realize every time I say that, I would mean at least one of five different things. The dominant themes, though, are public/private and cloud/non-cloud – traditional infrastructure versus not. The others are mixtures of software/platform and infrastructure services. Cloud is not a pure landscape at all.
Benchmark: It seems that, unless you’re a brand-new start-up, you’re going to have to end up with some sort of hybrid, because you’re not going to just abandon everything and go straight to the cloud.
Dennis Drogseth: Right. To actually use it well you have to deconstruct the enabling technologies included in the term and see what they really are. Again, cloud is a pallet of enabling technologies. It is not a "journey" to anything. It is not an end game. The notion of the goal being to move to the cloud or put something in the cloud is actually nonsensical, in spite of all the very real values of leveraging these technologies.
Benchmark: So assuming we’re pursuing it for the right strategic reasons, what did your research show about how the cloud is being incorporated into an enterprise’s overall IT management scheme?
Dennis Drogseth: We saw that having a cross-domain service management group was clearly an advantage, hands-down, in assimilating cloud technologies, and also in improving the overall credibility of IT.
And that typically means a group that’s governed either at the C level or the VP level, that has some roots in architecture, that works across domains, that knows how to talk to stakeholders in different groups within IT and across the business.
Benchmark: The study also looked at user experience management. What did you find?
Dennis Drogseth: User experience management brought benefits in virtually all areas for cloud. We correlated groups that had user experience management with those that achieved certain kinds of benefits with cloud, so it’s not a pure play, but it gives at least a feel of relative impact. And those groups with user experience management in place had categorically better results in terms of reducing costs from cloud.
They were one-and-a-half times more likely to reduce complexity of management through clouds and the use of cloud; 1.4 times more likely to reduce operational costs; 1.4 times more likely to accelerate deployment of existing services. Just as a few examples.
Benchmark: How widespread is the user experience approach? You’re correlating a lot of benefits to it, but how widespread is it among the enterprises that you looked at?
Dennis Drogseth: Among the respondents we looked at, only 28 percent had true user experience management, whereas 52 percent had strong service level management tools. My feeling for these numbers is that they’re low, especially when I contrast them with some of other EMA research data. If it were up to me to say cosmically what I think the adoption rate is, it’s probably more like 35 percent, about a third or a little higher.
If we sat down and said what do you think user experience management is or we explain it, that it could be observed, it could be synthetic, it could be instrumented at the end station, instrumented at the data center, etc., then once they go through the thought process, then they can decide well, that might really describe them. So I think 28 percent is a tad low but not dramatically so.
Benchmark: Isn’t one of the biggest challenges to overcome the fact that it’s difficult to achieve the same kind of visibility as when it’s all inside your house?
Dennis Drogseth: A lot of what we’re saying here is, if you have ways of measuring the efficiency of what you’re doing and then contrasting that to how cloud services are performing – having visibility into what’s going on – then you’re well along the road to optimizing your cloud investments. The notion that the cloud is supposed to be invisible is in the end a misnomer, especially for enterprise IT.
The enterprise is not saying fine, it’s invisible, I’ll make a leap of faith and hope it works. The enterprise is saying, how do we make cloud visible so we can trust it?
There are ways of being able to test user experience, both in the way that Keynote can, and through additional internal monitoring. Looking at end-station responsiveness of cloud services is one of the constants in optimizing for cloud.
Benchmark: Can you give us a sense of the cost and operational consequences of poor cloud performance – what it costs when it’s not working right?
Dennis Drogseth: Once again, cloud is this motley crew of things – in the end, it parses out to a series of separate results based on the types of service. Cloud is not a monolithic thing.
But as an example, I was talking to a group about downtime for financial transactions. It was a million dollars a minute. So you probably wouldn’t fully depend on an external cloud provider for that app.
For businesses with critical apps, our average estimate is $100 thousand a minute. But here’s the record – I remember the conversation. I was glad there was another analyst on the call because I almost passed out. It was for a financial institution, and their downtime cost was $14 million per minute. They were a clearinghouse for other financial institutions to do transactions, and all the trades lost and the volumes lost in a minute amounted to $14 million.
So they’re probably not an Amazon EC2 account, right?
Benchmark: It would be hard to imagine any service provider signing an SLA with those kinds of numbers in it. Are the groups you surveyed getting solid SLAs?
Dennis Drogseth: Of the group we surveyed, 70 percent had formalized service level agreements with external cloud service providers, which is quite interesting since many cloud service providers don’t offer anything beyond junk for SLAs.
And 65 percent have formalized operational-level agreements or SLAs, depending on what they call it, with internally supported groups.
Benchmark: What kind of best practices do you see or recommend in terms of measuring and monitoring cloud performance?
Dennis Drogseth: First of all, you should always know why you’re doing cloud. You should deconstruct it – it’s a set of technologies – and realize they’re a means to an end, all of them. And then figure out how they affect your broader service management objectives from a business perspective. Never go to cloud just because it’s a bright and shiny word. Many of these technologies are very valuable, but please don’t think of them as end games.
Number two, there should be some kind of central oversight that should involve a cross-domain group. It should have an executive view and there should be someone who’s getting their arms around all the chaos that could be associated with these various technologies, so you have a way of bringing them together to meaningfully benefit business objectives.
And then number three, is make cloud visible. User experience is probably the ultimate metric for all cloud services, because in the end you can tell if your initiative is improving performance or degrading it.
Service level is key. If you’re an enterprise and your cloud provider is not committing to service levels that you feel comfortable with, kick them out. The industry will change to a commodity set of service providers who don’t care about you. A more partner-oriented group that will start to give you visibility as you need it – that’s the group you want to work with.
In the end your products or applications as an IT organization – whether you harvest them through providers or whether you develop them, whether you deliver them partly through the cloud or not – will stand or fall on the success of the quality of your application services and your ability to deliver them cost-effectively and responsively.
If you give up those metrics just because you think you’re going to cloud, you’ve lost before you started.
About Dennis Drogseth
Dennis Drogseth is vice president at Enterprise Management Associates. He has over thirty years of experience in various aspects of marketing and business planning for service management solutions. Dennis supports EMA through leadership in Business Service Management (BSM), CMDB Systems, automation systems and service-centric financial optimization. He also works across practice areas to promote dialogs across critical areas of technology and market interdependencies. Prior to this, Dennis helped to build the network management practice area at EMA; before EMA, he worked at Cabetron and IBM. Dennis holds a B.A., magna cum laude, from Yale University.