Switching to outsourced IT

Senior managers should listen when IT departments float the possibility of buying IT infrastructure services. There's a good chance the IT department is either forward-looking or on the cusp of melting down.

Nowadays, ICT in business is less about wires, plugs, and computers, and much more about outcomes from applications. In the same way a really fast fibre network is useless without content flashing around it, computing is largely worthless unless the work it does makes your company faster, more reliable, and provides information that supports good decision making.

Focusing on outcomes rather than inputs is perhaps one big reason why businesses increasingly buy computing and infrastructure services from specialist providers. The opportunity to change IT from a management headache to contract-based service delivery minimises risk, capital cost, and the IT footprint. It just makes good sense.However, there are certain ICT elements intrinsically important to business points of difference. Outsourcing those is the same as relinquishing heart and soul. Better to focus on the boring bits of IT – the wires, operating systems and databases; the bits that many organisation users never really get down to and quite often cause the problems.

While some outsourcing decisions are driven by hard economic times, making cash retention a priority; other decisions reflect an acceptance that the business of computing isn't core business and all that's required is simply the outputs.

Whichever is the case, how does one find the right partner? Is outsourcing like buying any other service or long-term contractual relationship? No, it is not. Computing services directly impact on the operational heartbeat of the company, its business relationships, marketing and production. Service evaluation and selection must go under a microscope. Let's assume that all technicians are equally as competent as each other; that they all have access to the "smarts" needed to keep computers going; and that all data centers are built to be green, reliable, available and capable of hosting as many servers as required. It's not true, but it does help me to explain how I would approach buying an IT service.Your job is to ask the right questions.

Try these ones:

Q1: How many data centers underpin IT services?IT infrastructure services are much more than a network connected to a data center, even when it's a really big one. Bottom line: you don't have a secure utility computing services without a minimum of three local synchronised data centers offering tick-box services to get the level of protection your environment needs.

All business critical applications and data must have synchronised copies safeguarded somewhere else. That's your second data center. Then, if you're serious about business continuity, workloads of either data center must instantly switch to another location. That's data center number three. It's called cloud compliant utility infrastructure.

Q2: Are the data centers situated on solid ground?

Avoid flood plains, flight paths, tsunami wet-zones and earthquake zones. Data Centers must have easy access to power and communications networks – ideally from diverse routes.

Q3: How is service performance reported, and what happens when it's not up to scratch?

What metrics determine acceptable performance and are there penalties for poor service? Uptime guarantees are a given, but assurances must go deeper. The big issue facing buyers of virtualised utility capacity is over subscription. Some providers will over subscribe virtual infrastructure capacity, leaving customers exposed. Providers must guarantee the absolute minimum level of on-demand virtual machine units available.

Taking into account that you don't run everything 100 percent of time, you will generally get more performance than guaranteed minimum levels. But at the least every virtual machine unit should have another unit as the guaranteed minimum. Utility service providers who oversubscribe will arbitrarily specify a minimum, and when extra performance is required it will be robbed from someone else.

Q4: Can I speak to customers?

References are essential, depending on the size of the transaction, and illuminate gaps between what a supplier says and what they actually do.

Q5: Is this company focused on this type of activity, or is this an opportunistic leap?

Customers should be wary of so-called utility computing providers, of which many have little or no experience in the specialised infrastructure required to keep services afloat. Definition vagaries have sparked an IT industry land-grab, where every man and his dog with a server and Internet connection is suddenly in the business of IT infrastructure services.

The currently popular flavour – short-term rental of scalable servers and storage over the Internet – has been available in New Zealand for at least four years. As hype leapfrogs reality and inexperienced operators gloss over fundamentals, customers shouldn't be fooled by online providers who overnight change the signs hanging in their windows from ISP to utility computing services. They must ask to see what's happening out the back. The physics of delivery might be easier and cheaper, but it's the back-end people you have to worry about.

Q6: What does data center redundancy look like?  

Redundancy corner cutting is most often seen in the following areas: power supply to the data center/computer room; power distribution to the environment where the equipment's running; equipment connected to that power distribution; storage networks; communications network; routing equipment; perimeter network; ISP and Telco connections (are they going through diverse parts of the building). It's not cheap. Can you withstand a power outage?

"Yes, we're OK, because we've got a power generator," they'll reply. But generators don't kick in for 12 seconds. UPSs keep computers going, but air handlers stop. In blade centres that's an awful lot of hot air. Twelve seconds worth can take 12 hours to dissipate. In our own case, we addressed the problem by developing super cooled Type-R eco-pods. Q7: Is the agreed price what I actually end up paying?  

Test the proposal with what ifs, such as: What is the incremental cost of upgrading software? What if the platform can't perform all the tasks we expected it to do? Q8: Is there a roadmap plotting data center upgrades and future technologies?

Simply focusing on current day capability isn't enough. Data Center service providers should have a roadmap targeting emerging technologies and upgrade paths.  When all of that is resolved, the question of value has to be addressed. Where does value come from? Dropping staff? Reduced dependence on company individuals? Contracts for service versus promises from internal departments? Improved performance? More IT flexibility? Moving staff to more productive roles?