There is, of course, Gartner’s famous Hype Cycle which plots out the early stages of significant developments in IT technology, and it is fair to suggest that the Cloud is still very much in the early stages of being hyped up as the answer to everything. So it is often beneficial to consider a contrary view, such as that expressed by Jon Temple (JT), CEO of nlyte Software in conversation with BusinessCloud9's infrastructure editor Martin Banks (MB).
The company specialises in datacentre performance management tools, particularly those that manage the space, cooling and energy consumption of datacentres. From that position, he can see both the hype and potential of the Cloud and their possible impact on established businesses wrestling with the question of when or indeed whether to move to a Cloud architecture. Some, he thinks, may be falling for a misconception of what the Cloud actually is.
MB: You have been quoted as suggesting that the Cloud is just a fad. On what basis do you think that?
JT: Well, that’s a great question. As a company, nlyte benefits from the migration to Cloud Computing as it creates more complexity in a distributed computing environment, but puts a lot of energy back into the datacentre. So Cloud is a trend that we’re tracking. But from a business application perspective, save for some well-documented examples such as Salesforce.com, what we see is a lot of companies that try to go down the Cloud route but are struggling to break through as real stand-alone businesses. So when I say it’s a fad and not a trend yet what I mean is that, save for a few well-publicised companies, we don’t yet see the wholesale adoption of Cloud Computing for the implementation of business applications.
MB: But the growth rate of Cloud service providers is growing rapidly – 25-30% and more. Surely that does constitute a trend?
JT: Do you mean co-location is growing or Cloud Computing?
MB: Both, they intertwine. There is a perception problem where some companies say, `let’s build a private Cloud ’ when what they really mean is let’s build another datacentre that’s ours. That isn’t really the Cloud at all. It may use Cloud architectures but is not using service provider resources and saving the costs of the datacentre.
JT: Exactly, a private Cloud is not in my mind delivering any of the benefits of the Cloud . All you are doing is building another datacentre. They may be using Cloud architectures, but they don’t get any of the other advantages such as no upfront capital expenditure. The movement to outsourcing and the use of co-location facilities is only in part being driven by Cloud computing. People don’t necessarily just want to have their apps hosted over the internet, they want to get out of the business of running datacentres. That moves the problem from inside the company to outside, and given that we specialise in the optimal efficiency of power, space and cooling means that that market is a growing trend.
MB: I assume that while you have a customer base that includes individual businesses, the big opportunity now is providing services to the Cloud service providers. You should be an ideal partner for them.
JT: yes. I’d like to suggest that nlyte is in a no-lose situation. If the trend doesn’t take off then datacentres remain in the business and we can help end users reduce their energy consumption and make better use of their space. If the trend carries on we benefit from that as well. It is a trend we are monitoring closely.
MB: When you call Cloud a fad you are thinking about when companies talk about private Cloud they really mean having two or more datacentres that they physically own. They may use Cloud architectures and they may say they have Cloud , but they haven’t really. Is that fair?
JT: That is fair. It’s a fad for two reasons. There is a big difference between what is being branded private Cloud and what I would call a real internet Cloud such as Salesforce.com, where you are off-loading all the capital expenditure and work to someone else and taking advantage of the low cost delivery mechanism. If you’re building your own internet-based architecture, that is all you get, but you don’t lose the high capital expenditure. I also think it is a fad because it is not suitable for every application domain. It depends on the application whether users are prepared to let their data run outside the firewall.
MB: When it comes to datacentre performance what do you provide that can’t be provided by proper use of existing tools?
JT: You need to look at what we don’t do. We are 100% complementary to the big four datacentre infrastructure vendors, HP, IBM, CA and BMC. We’re not focussing on optimisation of network or systems management where those guys do a great job. We focus on the physical aspects of the datacentre: space, power and cooling. We use our intelligent capacity planning capabilities to help datacentre users reduce their operating expenses by up to 20% a year. We can also preserve the life of an existing datacentre by up to five years. And on a day-to-day basis we can reduce the time taken to deploy assets in the datacentre by up to 50%. So we’re all about helping existing datacentres get more out of what they’ve got, and if you’re building a new datacentre we can help you build it more efficiently on day one. Lastly, we can help optimisation and migration in consolidation projects.
MB: So is the primary long term target being part of datacentre design processes?
JT: Yes, it is one target market, but we also do a lot of work with existing users. For example, one capability with nlyte is modelling changes to datacentres before they actually happen so you can play `what if’ games. There is a spectrum of user needs here. At one end is when they have 50 new servers turn up and the question is where are they going to be put in the datacentre? The application has a materials catalogue, plus it builds a digital map of the datacentre estate. So we can map these two together to get the optimal location in terms of space, power and cooling. And that is not just first time, but every time there is any change to the datacentre. At the other end of the spectrum there is consolidation, where three or four datacentres come into two. Users need to model that effectively before they start moving assets around.
MB: So this could apply to the service provider market where the need to expand and restructure their resources will be a regular occurrence?
JT: Exactly, the status quo in a typical datacentre is about five minutes, because there is constant decommissioning of old equipment, new equipment coming in….
MB: Do you have tools that can identify systems that need to be decommissioned or can be decommissioned and plan out that process? It is often the case that users aren’t sure about decommissioning systems, so don’t just in case?
JT: This is an unbelievably important subject. This is one of the biggest values we can provide. Did you know that 15-20% of all the assets in a normal medium-sized datacentre are what I would call `orphans’. They’re sitting on the network, powered up, but not doing anything. Part of our Datacentre Performance Management Suite is integrated automatic agent list discovery, which automatically catalogues all the assets on the network. We can then relate those assets to the business processes to identify whether they are doing anything. Just identifying the orphans and turning them off can save a large amount of energy and space. If they have just low utilisation we can also model where that workload can be redeployed in the datacentre. When you consider that a typical datacentre now costs between $2,000-$3,000 per square foot, anything you can do to extend the life of your datacentre by a few years give a massive financial advantage when the capital expenditure is some $200m. It also reduces opex through the better placement of assets.
MB: Are there any limits to the assets you can handle? You mentioned IBM, BMC and CA, so presumably this includes mainframe systems.
JT: The architecture of the servers is irrelevant to us. In fact the more heterogeneous the environment is the more we thrive on it. In many datacentres today the users don’t know what assets they have, where they are, how much power and space they’re consuming. More importantly, they don’t know when they’re going to run out of those assets. So we provide predictive analytics that projects historical data out into the future. This includes modelling in the energy resources right back to the grid. Right now, many IT managers don’t know whether their resources will run out tomorrow, next week or next year.
Post new Comment