Many businesses find that, as they grow, the IT system fails to grow with them. They begin with a couple of PCs networked together by the director who’s always fiddling with these things at home, they grow unmanaged and then it’s time to get the whole thing under control. Or it’s done professionally, in an organised fashion – and then it turns out you’ve grown to a larger size than your existing systems were designed to support.
Years ago, the only option would have been to get a consultant in to redesign and re-engineer your system. This is still a good starting point, although the systems in whose direction they are likely to point you have changed drastically. Specifically, they’ll use two phrases a lot: one is “outsource”, the other is “cloud”.
Ben Gladstone is the chief executive of the outsourced IT company Conosco, and specialises in setting up systems for smaller and growing businesses. He believes a lot of companies suffer from an abandonment of common sense when it comes to new systems. The overall specifications, he says, should include low cost, scalability, redundancy and a quick restore after a failure (“redundancy” in this case meaning that one or two of the computers should duplicate the work of another, so if one component goes wrong the spare can kick in immediately). “For a small business, these are hard to achieve at a realistic cost,” he says. “They really demand replicated servers in separate locations – unless you use hosted solutions, also known as ‘the cloud’. Google Apps is the leading contender for email – and you also get Google Docs, which might replace your MS Office.”
The wiser upgrader takes all this into account, and then doesn’t do it all at once, according to Joe White, managing director of Gandi.net, which specialises in cloud services. The first thing to do, he suggests, is to make sure your supplier understands your business and preferably has worked with similar organisations. They can then get an idea of what you’re going to need, but be careful. “You tend to find that companies use about 10-12 per cent of their servers’ capacity,” he says. This doesn’t make economic or ecological sense.
Gandi’s answer, unsurprisingly, is to let them virtualise your systems. Virtualising means making one system function as more than one physical computer by putting it all into software. This can be done across the globe.
“Geo-locating can be useful because you can have different parts of your website located across different parts of the world,” says White – so if one part of your website attracts primar-ily an American audience, that part can be located on servers in America, for example. This won’t make any difference to you as the site owner.
Beyond a certain size of business, you’ll need to look at a data centre, whether it’s in-house or outsourced to the cloud.
Chris Smith is sales and marketing director at on365, which specialises in this sort of infrastructure. He observes that many companies put the server computers into the nearest available space – which doesn’t always work.
“The selected office space takes away ‘people space’, so is often forced upon the team as being the space that is the least attractive for office space,” he says. “It may have general building water or waste pipes running over it, refrigeration pipe-work, condensate drains etc, all passing overhead.” These all put the IT system and the data residing on it at risk.
Mitchell Feldman, managing director of the Internet Group, who focuses mainly on sales and marketing, adds another dimension. “Align IT requirements with future business plans. It is essential that a company’s business plan acts as a blueprint for the design of the new IT system. An IT infrastructure should be developed to support the company’s future growth plans, not work as an unrelated entity.
“Conduct a thorough risk analysis – consider what role IT plays within the business, and carry out an impact analysis to fully understand and prepare for the ramifications of any system failures.” Note that, whatever any supplier tells you, the system will fall over at some stage – and there needs to be provision for what happens then, and just how often it’s reasonable to put up with this in the contract. This is called the Service Level Agreement (SLA) – and any SLA that specifies a system will work 100 per cent of the time isn’t realistic.
There are a few practicalities to bear in mind. Conosco’s Gladstone suggests avoiding customisations, as they can be costly to fix when they go wrong. Look for next-day on-site support whenever there’s a problem, and have that written into the contract. Gandi.net’s White suggests avoiding upgrading everything at once – replace things when they break and not before.
A related area is knowing not only why you’re replacing your existing IT but why you’re opting for whatever is the new version.
Alastair Williams is the data management practice lead at Centiq. Although he acknowledges the cloud can have huge benefits, there are caveats. “One such area is around unstructured data management. When combined with intelligent archiving policies and process management, storing and distributing business documents within the cloud can have significant benefits,” he says. “The danger, however, is simply moving data to a new location without addressing the ‘why’ behind it. Ignoring this will simply move escalating costs to a new budget line.”