To build or to borrow?

Written by Phil Alsop, Editor, DCS Europe Published 2017-08-17 08:20:50

Us Brits like to buy our own house, rather than rent. But, when it comes to cars, for example, companies tend to lease vehicles for a fixed term, at a fixed monthly cost, preferring the budgetary and service certainty of this option, rather than having to manage and maintain a whole fleet of vehicles. More recently, private individuals have started going down the same route – paying a deposit for a vehicle, making relatively small monthly payments for, typically, three years, and then choosing to hand back the car in exchange for the latest model, rather than make the final, large ownership payment. The downside of this approach is that you never get to own a vehicle, and you can end up paying significantly more for your transport by leasing it rather than making a capital purchase.

The upside? You get a brand new car every three years, someone else looks after all the maintenance, and you have a fixed monthly cost around which to budget – rather than the vagaries of random breakdowns and expenses if you own the vehicle yourself.
Substitute the words ‘vehicle’ and ‘car’ for ‘data centre’, and in broad terms, the same arguments hold good. If money is no object, and you must own your own data centre, then all good and well. Indeed, over the likely lifetime of the facility, potentially you’ll save a lot of money in terms of not having to pay a monthly rental fee (i.e. at some stage in the future, the overall cost of renting the facility will exceed the cost of building it), and you have the security of knowing that you control the data centre.

Furthermore, your data centre specification might be outside the remit of the colocation providers, so you can only obtain exactly what you want by building it yourself. And, no matter what the colocation providers might tell you, you like the idea of being able to talk to your own staff when problems occur, rather than to some third party who just maybe doesn’t appear to have the same level of customer service as your in-house team.

Of course, the cost of building, owning and operating your own data facility is not insignificant, and then there’s the cost of maintenance – planned or not – and perhaps the slight, nagging concern that you don’t have enough in-house expertise to feel comfortable that, no matter what issues might arise in the data centre, your own staff can deal with it swiftly and competently. Search the internet for ‘data centres and outages’ (or ‘disasters’), and you’ll discover, or be reminded of, one or two recent news headlines where even global organisations seem to have struggled to maintain a reliable IT delivery service, courtesy of problems with their own data centres!

Always assuming that such issues don’t cause any sleepless nights, there’s the less easy-to-dismiss debate to be had around the data centre infrastructure refresh rate. You’ve stretched the budget to build your own facility. It’s costing a pretty penny to operate and maintain and then you’re told that, in order to take advantage of technology advances in say, UPS or cooling technology, you need to replace your existing plant with some new kit. Or, maybe, the overall design of the data centre needs to be updated – modular, hot/cold aisle containment options need to be considered, rather than having one huge, open data hall.

And we haven’t even thought about the emergence of the edge. So, the huge centralised data centre you built two or three years ago now needs to be significantly downsized, and you need to build, operate, and maintain several edge data centre facilities, to ensure that data that is created locally can, where necessary, be processed and stored locally to avoid issues to do with, maybe, latency and/or security. Or, centrally stored data needs to be moved to more local, strategic locations on an occasional basis, to ensure speed and quality of delivery.

What about flexibility? Your own data centre is built to house your predicted IT needs for, say, the next four years. But what if this growth rate is over-optimistic? You’ll be paying for lots of empty space. Take it to the other extreme, and two years later, your data centre is full. Does your budget allow for another data centre to be built? And can it be built in time to meet the ever-growing demand?

Much more likely is that your overall data centre requirement will fluctuate. Almost regardless of the industry sector in which you work, there will be significant workload peaks and troughs. Everyone knows just how busy the retail sector is in the run up to Christmas. So, do you have a data centre built to cope with this peak capacity, and, therefore, have infrastructure that sits idle for ten months of the year? And the oil and gas and other research-hungry industry sectors need to run some serious data processing/analytics programmes, but not all the year round. Again, do you scale up for the peak demand, and then leave the infrastructure unused for large periods of time?

And then there’s the start-up scenario. How many new companies can actually afford to build their own data centre facility long before they have any revenue streams? Imagine how that conversation would go with the bank manager (or, more likely, the computer programme tasked with deciding your fate!). Clearly, if you’re thinking big, and IT is a major, critical part of your business, then building and owning your own data centre and IT infrastructure could well matter, but for most size and shape of business it’s most likely out of reach when it comes to start-up mode.

‘Hybrid’ is a word that’s become mainstream in the IT world when it comes to discussing one of the major trends of the past few years: Cloud. There seems to be a general agreement that the best Cloud model for most companies to adopt is the Hybrid Cloud – some use of in-house data centre/IT resources to create a Private Cloud, and some use of third party, Public Cloud infrastructure. Public Cloud use might cover the IT/data centre ‘basics’ or commodity items, where it’s difficult to add value in-house, and bursting – making use of extra, third party compute/infrastructure resources at a time of extra demand – is another common use case.

So, maybe this same hybrid model could apply to the data centre facility dilemma? Possibly. After all, you might want to sweat your existing data centre assets just a little longer to keep the accountants happy. But does your in-house data centre offer a truly scalable, affordable, state-of-the-art, economy of scale-benefitting, resilient environment? Indeed, can it ever?

A colocation provider can offer a customer just a couple of racks to start with, allowing them to expand as and when required (either for long-term business growth, or seasonal peaks). These first few racks can be located in a modern data centre facility, where the infrastructure is refreshed regularly, where the security and resilience are world-class, and where you obtain a significant economy of scale, because there are lots of other customers like you, all with their few racks or cabinets, building out into a well-used, energy-efficient, truly optimised data centre facility.

And if all of the above is not enough to convince you to have a serious think about what’s sensible when it comes to data centre ownership. Ask yourself why your organisation doesn’t own its own power station, and telecoms exchange? If you do, then maybe data centre ownership is for you. Otherwise, maybe it’s time to overcome that psychological barrier and recognise that paying a fixed monthly fee for a defined, reliable data centre resource, that would be well out of your price range if you had to buy the whole thing, does make sense.