There isn’t much about Tom Sanzone that bespeaks drama. The CIO of Credit Suisse is direct, meticulous and practical, and it doesn’t seem as if he’d suffer fools gladly — an impression partly informed by his New York accent, nearly shaven head and confident demeanour. But ask him what virtualisation has helped him deliver to Credit Suisse, and you’ll get a dramatic answer: “Tremendous results”. He’s not just talking about savings reaped from data centre consolidation — which was what the first wave of virtualisation projects was all about. Sanzone and other leading CIOs are taking virtualisation to the next level. They’re using it to become the fast, flexible business partners that CEOs have always wanted.
Balancing data centre loads
One example: Virtualisation has radically changed the way Sanzone’s IT group delivers computing power to the company’s application-hungry line of business units.
The old process of allocating a server box for a business unit — including purchasing, provisioning and configuring the hardware — took weeks or months. Now, Sanzone says, with his growing number of virtualised servers, “it can be done in one day.
“And we want to move to be even quicker.”
This, he notes, is the kind of benefit his business partners understand loud and clear. “They can see the impact of this type of technology and what it can mean to the business,” he says. Such as? “Such as a quicker time to market for the products they need, competitive advantage and driving revenue growth.”
Sprung out from server-farm savings and spreading to Credit Suisse’s four core business units, virtualisation has become a central piece of the US$67 billion financial services company’s future.
Stephen Elliot, a research manager in the enterprise systems group of IDC, says this piece of technology can be vital for CIOs who are struggling with the business side’s expectations. “Virtualisation is an opportunity to strike more of a partnership with the business, because you can drive out a more agile infrastructure and meet those [business] requirements faster,” Elliot says.
CIOs and analysts agree that IT departments need to move beyond simple “data centre consolidation” thinking. Virtualisation can enable dramatically faster provisioning of equipment and computing resources, better chargeback systems and stronger disaster recovery capabilities. “That’s when a CIO moves from tactical to strategic,” Elliot says.
Like Sanzone, many CIOs realise virtualisation is much more than a cost-saving tool. But almost every CIO’s journey to a virtualised environment will start with a data centre consolidation project — and big savings that provide a pleasant starting point with the business side.
Virtualisation technology from vendors such as VMWare (which owns about 80 per cent of the Fortune 1000 market) acts as an “abstraction layer” between physical server boxes and the operating systems and software running on them. This means that up to 30 virtual servers can be housed on just one physical server, though that number varies widely from data centre to data centre (depending on how resource-hungry a company’s applications are, for example). CIOs can then create a virtual pool of computing power, which reduces the need for a huge number of servers and allows for much better CPU utilisation rates (which typically run anywhere from 5 per cent to 15 per cent) in servers.
How savings add up
Out on the edge of the virtualisation frontier are IT chiefs like David Siles, CTO of the Kane County government in Illinois. In addition to his server virtualisation, data consolidation and disaster recovery successes, which he started in 2005, Siles has embarked on an ambitious desktop virtualisation project.
Siles is about a quarter of the way through deploying VMware’s virtual desktop infrastructure to the county’s user base, which accounts for 2000 PCs.
Those users who have started to switch over, including administrative staff, politicians and sheriffs, use thin-client terminals that are hosted back in the data centre on virtual machines.
Although Siles says it’s still early in the changeover to calculate exact savings, he will eventually be able to cut maintenance, hardware and service costs because everything is controlled inside the data centre. For example, each terminal cost him US$600. But since he has been able to get 50 to 60 hosted desktops on each four-processor server, he has been able to cut in half each terminal’s cost. And, he notes, terminals boast an eight-year lifecycle, rather than a four-year one for a PC. Those thin clients — especially the 55 mobile units — now pose less of a security concern. “We had a couple of laptops stolen out of police cars,” Siles says. “Now [with the virtualised thin clients], you essentially just lose a dumb terminal.”
At Credit Suisse, Sanzone will remove more than 2500 of the company’s servers (which total more than 20,000) by year’s end. In the development environment, he has been able to realise a 20-to-one ratio for consolidating servers, meaning he consolidated 20 physical servers onto one new server. He has also upped utilisation rates on those servers from some as low as 7 per cent up to 40 per cent.
In turn, he expects to save millions on reductions in maintenance, hardware capital, networking, migration, real estate, and data centre power and cooling costs. Credit Suisse is on track to reduce its data centre power consumption by one megawatt by year’s end — all courtesy of virtualisation. (Recent virtualisation research from Aberdeen Group shows that 83 per cent of “best in class” companies have increased their server utilisation rates, 57 per cent have decreased capital costs, and 52 per cent have reduced staffing overheads.)
According to IDC, 75 per cent of companies with 1000 or more employees are currently utilising virtualisation. And more than half of the respondents to the Aberdeen Group survey on virtualisation were using it in their production environments (meaning, everyday user applications), “a clear sign that end users trust the technology is mature enough for mission-critical systems”.
Data centre consolidation is a critical first step toward unlocking virtualisation’s potential. It demonstrates to the business that IT can run more cost-effectively — slashing server costs plus tackling expensive power, cooling and real estate expenses.
That first win and proof of concept with server consolidation invariably lead to more virtualisation conversations, say CIOs and analysts. CIOs should seize on these second-round discussions to expand upon how virtualisation can provide greater flexibility and speed in provisioning computing power to lines of business (LOBs), allow for greater utilisation of idle CPUs, speed up business continuity and disaster recovery operations, and improve application and system availability.
However, experts and practitioners say you should take it slow and keep it simple. That’s because even many who work inside IT still have difficulty understanding all of virtualisation’s complex terminology — just imagine how your CFO or marketing VP will feel if you start discussing hypervisors and CPUs. (Hint: Eyes glazing over, furious typing on BlackBerry.)
In fact, IDC’s Elliot says he has talked to some CIOs who have hidden the fact they have virtualised their servers (and realised cost savings) altogether. “They don’t want to spook anybody with tech jargon,” Elliot says. In a rather subtle way, he says, it’s conceivable that those CIOs who don’t inform the LOBs about their virtualised savings are actually able to make a “profit” on those services because of the virtualised savings. “If you can get away with that,” Elliot cautions.
This sneaking around is, in part, a response to the fact that some LOB chiefs have become ‘married’ to their servers. These executives are “server huggers” and they can be very territorial about their boxes. But it’s those server huggers, and their demands for loads of computing power for their applications, who are one of the chief causes of server sprawl in the CIOs’ data centres.
Rewriting recovery rules
For a growing number of CIOs, virtualisation has become a valuable tool in being able to quickly restart business operations and applications in times of man-made or natural disaster, and keep IT service interruptions (both planned and unplanned system maintenance) to a minimum.
Disaster recovery and backup operations have garnered much of the attention, simply because the time and cost savings CIOs can deliver to the business can be almost as dramatic as a flood or fire.
Chevy Chase Bank’s VP of IS Matt Wilson is currently migrating from a backup process that includes having to manage loads of backup tape. He says the new virtualised system reduces a 20-hour recovery process per Windows server to 15 minutes for each of his virtualised blade servers.
(Wilson has already been able to consolidate and virtualise his development group’s 100 servers to 14 blades, which amounts to a seven-to-one ratio of physical server to virtual servers; by the end of 2007, he will have moved 160 production servers to two blade centres.) He notes that a big advantage for the bank has been the speed at which his group can recover a virtual machine after a disaster.
Wilson’s now using a third-party company as his disaster recovery centre, and his virtualised infrastructure has been strategic to the bank’s operations.
First, he has been able to cut his monthly disaster recovery costs because he doesn’t have to own a duplicate set of hardware and all of its associated costs — the power and cooling costs that come from all of those idling CPUs just waiting for a disaster.
In addition, because it’s easy to replicate the contents, Wilson doesn’t need staffers to ensure the data between the two locations is in sync and that configuration changes and updates have been made in both locations, which can be incredibly complex due to dissimilarities in hardware types at the locations. “It’s a big time-saver,” he says. When asked what the value of his efforts is to the business side, his answer is simple yet critical: “We can quickly recover business applications — that’s the value.”
Sidebar: Businesses to save millions from virtualisation
Infrastructure virtualisation will be the dominate technology in data centres within the next few years, thanks to the convergence of three significant factors in the global economy, according to analyst firm Butler Group.
Virtualisation is the practice of running a layer of software on a server that allows multiple operating systems and environments to run on the same piece of hardware as if they were separate physical servers. Virtualisation allows computing resources to be used more efficiently and also allows much greater flexibility in managing and allocating these resources.
Many vendors are touting the benefits of virtualisation, and it is perhaps no surprise then that Butler thinks the technology is set to dominate the data center.
The three factors influencing this, says Butler in its report, Infrastructure Virtualisation, are the need for organisations to reduce their energy consumption and carbon footprint; the increasing need of being able to respond quicker to market opportunities; and the drive towards automation in order to reduce operational costs.
Butler warns however, that infrastructure virtualisation requires a significant change in an organisation's culture from both an IT and business perspective. Companies have to move away from their "siloed business unit autonomy position towards a virtual business process-driven architecture”. This is supported by the adoption of infrastructure virtualisation.
It believes an organisation running 250 dual-core servers can gain a cost saving of £2 million over three years from server consolidation alone. Furthermore, power saving in the order of £78,000 per 1000 PCs per year can be realized by moving from a full desktop PC infrastructure, to a server-hosted desktop virtualisation solution.
Of course, moving from physical to virtual might appear to be straightforward on paper, but the migration of production systems into a virtual environment presents a very real set of challenges, says Butler.
It also thinks storage virtualisation can help organisations to manage their storage resources more efficiently in order to achieve higher utilization rates.
On average, organisations can save £4000 per 1000 helpdesk calls per month through the promotion of user self-service and a reduction in application-related help desk calls, Butler estimates. But it warns that IT managers and enterprise architects must understand and appreciate the key elements and concepts of network virtualisation if they are to run successful IT virtualisation projects.
Finally, Butler believes that even greater benefit can be obtained through the strategic adoption of virtualisation in the data center. and that infrastructure running cost-reductions in the order of 40 to per cent are certainly attainable in many cases.
By Tom Jowitt, Techworld.com
Join the CIO New Zealand group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.