Moving to the Cloud - How effective a panacea is it?

All the talk in technology management these days is about the cloud. That’s been the case for some time, and I don’t claim that I’m saying anything revolutionary here. But I do think it’s worth talking about whether it’s really the cure to all ills.

Broadly speaking I would argue that businesses should consume their applications and IT services in the following preference order:

  1. Software as a Service

  2. Platform as a Service

  3. Infrastructure as a Service

  4. On premise

This depends on a number of factors of course - compliance, availability, legacy requirements, user tolerance… there are always reasons not to do things in line with the theory.

Infrastructure on premise is a huge cost, usually capitalised, which transfers debt onto future years, sucking away future profitability. For that reason cloud services always look attractive in principle. But when you get into the weeds you find that returns are often much longer than hoped.

The big catch that no one likes to talk about: ‘cloud’ still means infrastructure. It’s someone else’s infrastructure, and that someone else has taken the capital hit so you can keep the costs in the OpEx column. But the costs are not that much different, particularly if you’re stuck in the IaaS layer. After all, your provider had to buy capacity as an investment, capitalise it over a similar period that you would have, and then recoup the accrued debt, plus make some profit. The only way to do that is to charge you for it.

So, if you’re thinking you’ll save money by shutting down a VM in your data centre and spinning one up in a cloud platform, think again. You might save some money eventually, but it will take longer than most boards are willing to wait.

So we need to look to Platform-as-a-Service wherever we can. Not all applications will run in this environment, although that situation is getting better all the time. The platform providers are also getting better at offering more forgiving solutions that will support applications that were not supported a short while ago. Sharing and leveraging compute resource and clever database design can keep costs down but can also create complexity for system administrators that needs to be taken into account.

Another thing that cloud service salespeople never tell you. Cloud platform and infrastructure services need as much management as on-premise solutions. Platform providers are not psychic, they won’t spontaneously create an environment that meets your every business need, and they won’t sit there tuning it for you either. Of course, you can outsource your cloud management - there are lots of very good companies who will do that for you - but it’s still a cost that doesn’t go away just because you’re saying cloud.

Ultimately, the ideal is Software-as-a-Service. Like IaaS and PaaS, SaaS still needs to be managed, but with different skills which overlap heavily with skills you’ll need however you take your applications. The problem is that SaaS is often much more expensive per user than other application offerings, at least in the obvious and apparent cost.

Imagine you own a business that has an on-premise CRM with 100 users. You bought the licences 4 years ago at a cost of £100,000 but that cost is long since sunk and depreciated to nothing. The server is a bit wobbly, but it works when you need it. You bought that at the same time, again depreciated to nothing. And every year you pay 18% in maintenance fees. Your annual run cost is £18,000 plus a bit of time from an IT technician whose costs are probably sunk in another cost centre anyway while they support this as a side job.

Now, your provider has a great idea - let’s get you on our Software as a Service offering! It’s so much better. They of course have to provide infrastructure at a cost of £15,000 a year, plus support and maintenance, £10,000 a year, and their bandwidth costs which run to another £10,000. They will discount your software support by 50%, making it £9,000 a year.

So for this new service, your per-user cost has gone from an apparent £180 to £440. And of course you have a migration project on your hands.

Cloud IS a great solution to many problems, and it IS cost effective in the long run. But it is not going to solve a cash hole in a single fiscal year and you shouldn’t be doing it if that’s your motivation. Do it because of the many benefits, which I will write about in another post.

Nick Ellis