The path to efficient and flexible IT systems starts with an understanding of the business and the systems you use today
Optimising the datacentre is a process that goes far beyond technology, and can have implications for the way you run your business. As we outline in our Strategy feature (pg.10), it may involve a rethink about how IT services are delivered within your organisation. But even if you don’t undertake any radical changes in that area, ensuring a successful optimisation project means having a clear understanding of the business. So the first step is to define your aims.
“Not every organisation is driven by the same requirements from the business, let alone the IT function, so the benefits are very organisation-specific,” says Terry Walby, datacentre solutions director for Computacenter Services. “One of the things we engage in very early in an optimisation programme is simply getting to understand the customer – what their business drivers are, what their circumstances are, what they’re trying to achieve, and what are the pressures on the IT department.”
There may be any number of drivers for an optimisation programme. You might need to free-up datacentre space. Perhaps you need to increase IT service levels to your internal customers, or increase the flexibility in the way they use your services. Or you might simply want to provide more visibility to your stakeholders in terms of how their investment in IT is being used. These strategic or tactical incentives may be accompanied by a range of cost drivers, too, including driving down maintenance and support costs and freeing capital expenditure to use on more core business activity. In that case, you have to know where costs are coming from before you start.
“Depending on which of those are true will determine what your approach will be,” says Colin Bradford, datacentre practice leader at Computacenter Services. “Then you’ll know whether it’s about refining technologies that will allow you do carry out consolidation, or whether it’s using appliances that will allow you to substantially reduce the number of servers. You might want to invest in tools that will give you a greater appreciation of the service level that’s being provided. And where it’s about cost, you may even want to look at out-tasking particular support functions because you don’t necessarily have the expertise in-house to manage these specific new technologies.”
He adds: “An optimised IT infrastructure is about right-sizing every aspect, including your investment in software licences and hardware, the size of your support team and the tools and processes they use. The whole point is to look at things differently.”
Process of discovery
The next step sounds simple but may hold some surprises. It’s the ‘discovery’ process, which means auditing your current IT systems to work out what you’ve got and how it’s performing.
“This doesn’t necessarily mean full-blown benchmarking,” says Zahid Din, consultancy technology leader at Computacenter Services. “You just need to get everything into a single, consolidated view. It’s amazing the number of projects we undertake where people declare they have a certain number of servers or applications, and then the discovery process reveals as many as 25 per cent or even 50 per cent more.”
As Computacenter Services has worked with a wide range of customers, of different sizes and shapes, in different stages of maturity and across a variety of sectors, it has experience in this discovery process. “We’re able to carry out a highly pragmatic exercise,” says Din, “so that, even if customers have been quite good at rationalising their systems or optimising their datacentres, we will usually find areas for improvements – for example, by using more high-density technology, or leveraging tools and third-party capability for additional cost savings.”
Interpreting the data
Understanding the results of this discovery process is obviously key to highlighting and prioritising the areas that need attention and those that will yield the biggest or most immediate returns.
“This is where experience really counts,” says Din. “Having been through this so many times before, we’re able to say, ‘we know you can consolidate those 10 or 20 applications on to this single, physical box, it will scale, it will deliver these benefits and we’ll show you how’.”
The details are typically thrashed out using a workshop to clarify the customer’s business and technical requirements. This casts light on some of the issues that have been brought up during the discovery process as the users and IT department may have specific reasons why things are done in a certain way, and that affects the programme as it goes forward. Key to this is building an understanding of the application interdependencies and mapping them to business processes – since this is critical in building a picture of the optimal co-existence between these business applications and processes, and the underlying architecture.
So too, is proof of concept. Once the architecture for the optimised systems has been designed, it is essential to be able to test it without impacting the current, live systems. This is a critical part of the de-risking process – precisely the function of Computacenter Services’ Solutions Centre (see pg.05). Here, the customer’s environment is replicated so that the new optimised technology, and the change process itself, can be tested and documented. Computacenter Services and the customer, working in partnership, use this to understand all the parameters and complexities involved, and discover the challenges they are likely to face as well as how to recover from issues if they occur. This testing and proving process is important because it would be wrong to underestimate the impact of a large technology change programme.
“If it was trivial, you would have done it a long time ago,” says Din. “There is some effort required, some commitment to the potential for gains is needed, but nevertheless it can be done in a very non-disruptive way.”
One approach is to do it piecemeal. It’s quite feasible to take one aspect of optimisation and use that to demonstrate the possible benefits. You can use this to win buy-in from the senior management – something you’re going to need if you’re also asking for changes in things like accounting activities, as discussed in the Strategy feature.
Showing how optimisation might improve people’s lives – or the bottom line for lines of business – is also good PR for when the full programme is underway, as there is inevitably going to be some disruption, as each component may involve some downtime.
“There has to be a programmatic view,” says Bradford. “That involves a lot of detailed planning, a lot of structured thought and a lot of high-level sponsorship. And yet the whole reason people want to undertake these optimisation programmes is that they want some short-term or immediate benefits, and they don’t get those from programmatic views. So that’s a challenge. It needs a balance of pragmatism, which is about delivering real tactical benefits – such as quickly freeing up datacentre floorspace, quickly consolidating servers, quickly building a virtualised storage environment – and a strategic view which allows you to deliver the longer term benefits and build the infrastructure for the future.”
Virtual servers
Of all the technologies and approaches deployed in an optimisation programme, server virtualisation can provide one of the easiest and fastest returns (see box). That’s because servers represent one of the areas of greatest inefficiency. Across the industry, average server utilisation is only 10-15 per cent, and Computacenter Services regularly finds examples as low as 3 per cent.
“At the root of the problem is the inherent lack of flexibility caused by the operating systems, applications and their associated data being bound to the hardware on which they are installed,” says Cadman Chui, VP marketing for PlateSpin. “This ‘software stack’ cannot be moved easily from one hardware platform to the next, or between physical and virtual infrastructures. This has resulted in datacentres that have many under-utilised (and some over-utilised) servers.”
The mushrooming of x86 servers is partly because they have such a low perceived cost – which is ironic considering that significant savings are available simply by consolidating them using virtualisation. With this technology, a single box running virtualisation software, such as VMware, is host to multiple virtual machines, each running its own server software as though sitting on a machine by itself. In addition to sharing the main system resources – processor and memory – the virtual servers also share any attached peripherals, including storage, making sure that every component is fully utilised.
Aggregation has similar benefits. This is where a number of components, such as blade servers or storage devices, are grouped so that they appear to act as one unit capable of running multiple, high-end applications. The key advantage is flexibility in scaling the system – up or down – to match your real, current requirements, rather than sizing it to meet some theoretical peak load. This technology also offers a level of resilience in that it is not dependent on a single device.
At a more general level, centralisation of resources such as storage brings with it all the management, security and right-sizing benefits we’ve discussed for datacentres in general. It’s surprising how often the supposedly ‘centralised’ environment of the datacentre actually contains a great number of storage systems attached directly to individual servers – an approach that is every bit as wasteful as the stovepiping of the servers themselves.
These are the kinds of technologies and approaches that can provide quick benefits. Extracting the full value of optimisation involves a little more work looking at how you do things. Once again, storage provides a good example.
“If you carry out some analysis around your data types, you’ll find you’re paying to store all kinds of information that shouldn’t be there, or which could be off-loaded to cheaper archive storage,” says Din. “People’s personal files often aren’t referred to after a while. Finance data can be safely archived after the year end. After three days you never look again at 90 per cent of your emails. So a carefully crafted, tiered storage policy can achieve quite significant savings.”
Many analysts predict that moving to a storage area network (SAN) architecture can pay for itself within a year. SAN and Network Attached Storage (NAS) systems allow you to meet growing or fluctuating storage demands while also providing efficient management capabilities.
More radical still is a move to a ‘pay as you go’ utility computing or capacity on demand infrastructure, where business users buy computing capacity when and as they need it from a shared service. This involves major cultural and architecture changes, but as we discuss in the Strategy feature, when matched to the appropriate accounting processes, it not only offers major cost savings and improvements in business agility, it also ensures buy-in from the business users, as they’re the ones seeing their IT bills reduced.
“By adopting a utility approach to computing services, organisations are able to vastly reduce the time to value from IT assets,” says Walby. “All businesses would love to only have to pay for what they are actually using. And for a commercial organisation, that can mean only paying for IT infrastructure when getting revenue from customers.”
Once the optimisation process is complete, there are still steps to take to ensure you continue to benefit. There are information and management tools – many of them automated – to help you monitor, measure and maintain efficiency levels. You can also make use of provisioning tools that mean you can repurpose servers on the fly when you need more processing power for one application, or to allow you to operate servers in one capacity during the day – for example, as a file and print server – but which overnight become a grid computing engine to do something completely different.
You don’t have to do all this yourself. In fact, given the skills required, you might find it more cost-effective and efficient to out-task aspects of your datacentre management. Computacenter Services, for example, has a number of managed service offerings that can work either with your own datacentre facilities or make use of outsourced hosting services.
“Rather than customers having to make large levels of capital expenditure up front, we can construct financial models that allow them to get quite complex infrastructure implemented quite quickly and they pay for that over a period of time,” says Bradford. “They also get to transfer the cost burden from capital expenditure to operational expenditure.”
While the desire to make cost savings is behind many optimisation programmes, with the right planning, technology and implementation, the real effects can be felt in the way that the revitalised IT services transform your business.
