When “Cloud” is not the only answer
I have many cloud deployments and have found “the cloud” to be a fantastic answer to many IT provisioning challenges.
But here’s the learning; “The Cloud” is not necessarily best value for everything.
I run the IT services for a Local Government authority where I am putting more and more services into the cloud. I have a blended cloud solution, with some services from a conventional data centre and some from the Cloud.
I have a dozen systems from cloud providers, either live or in the midst of a cloud migration project. These range from a benefits portal, to schools email, to theatre event management.
That represents a good portion of my enterprise systems. Some remain in our data centre, because there is not a compelling case to migrate them anywhere else. Like every government agency we have numerous legacy applications that run reliably and have done for many years. These remain in the data centre. Some systems in the cloud, some in the data centre; a somewhat pragmatic approach that the cloud purists seem to deride.
The building site
But more of the context. Like many local authorities, we recently agreed a plan to reduce our building footprint and in the process turn the data centre into a building site. So the existing data centre had to go.
“Fantastic opportunity to put everything into the cloud”, I thought. That would give the organisation many benefits and I’d be a hero of the digital fashionistas.
Imagine my surprise and disappointment when I looked at the actual numbers, not the theoretical white paper ones, and realised it will be considerably cheaper to build a replacement data centre than any other viable option.
“Moving everything to the cloud” has many possible meanings. It can mean moving each system to a system specific “true” cloud provider. So you end up with as many cloud providers as you have systems. To be cost effective, you procure as and when systems become available with cloud options and as they come up for renewal and refresh; which we did not have time for with the building project imminent.
Or Cloud can mean having someone else provide the whole data centre via “IaaS”. We used to call that “hosting”, “outsourcing” or “co-location”, but now the vendors need to call it cloud, for obvious marketing reasons.
The co-location option looked most desirable. I started with G-Cloud (Digital Market place), of course to see what the options and prices were. It is widely known that G-Cloud Framework prices are capped maximums and every supplier says they can beat their own prices on G-Cloud. So, of course I spoke directly to the suppliers as well.
I also spoke to most local public service organisations, across Health, Police and Local Government in the region, to see if they had capacity to host my entire data centre. None did, but I found two Local Authorities, one in Scotland and one in the South East who could. Their prices were very similar to the commercial co-location providers. The Crown Hosting Service had not at that time been launched.
Finally as a comparator, I also looked at the costs of building our own data centre.
The Economics of the Cloud
In our case it turned out to be substantially cheaper to buy a data centre, than to rent from the cloud. We own our own property and within that we identified a good space for a replacement data centre. This space could not be used for any other significant purpose. The major advantage of this location, is that it is on the same site as the major Council office buildings. So I don’t have to pay the data communications costs to connect all our enterprise users to our enterprise systems. This avoids a material cost.
Taking the data costs into consideration, on a like for like basis, co-locating in the cloud would have cost two to three times the cost of self provision.
I also host some systems for other authorities and am in discussions with commercial providers relating to hosting their services in my data centre – generating revenue for the authority, which would otherwise be lost.
The economic case for building our own data centre is extremely compelling. When the sales director of a leading cloud/co-location vendor tells me he cannot compete on costs against a self build with data costs, I know just how compelling that is.
Benefits of Cloud
Cloud is great when you are looking for elasticity, agility and resilience. So for those applications demanding that, it is worth paying extra for. But where we know our system workloads do not have significant fluctuation and where we can enable internet access for users, with reasonable resilience, why pay more?
Blended Cloud Architecture
So I emerge from this with a fairly common blended cloud architecture. Some on premise, some hosted elsewhere and some from true cloud.
Some people call the data centre a “private Cloud”. I don’t think that’s helpful, or even technically accurate. Over time I will, of course, migrate more services from Data Centre to cloud and procure more new services from the cloud. But I also will implement more services in the data centre, if that’s the location that works best.
I was amused by the press coverage as our data centre plans became known, but not altogether surprised. Fuelled by suppliers who wish to generate more revenue from cloud services (“cloud” has taken over from where “outsourcing” left off), some commentators seem to believe that there is only one answer to everything and that is cloud.
In my experience, there is never one answer to everything. There are innovations that need to be applied in judicious ways, where there is a business case.
To run a cost effective service, you need a certain amount of distance from believing in every “next big thing”. The business case is the king. Fashion is not a good enough reason.
I think of this in terms of the Gartner Hype Cycle. In the peak of inflated expectations, “enthusiastic and strong feeling around new forms of media and technology, in which people expect everything will be modified for the better,, is a common characteristic. We seem to be there with Cloud right now.
Moving from the peak of inflated expectations with “everything to the cloud”, to the plateau of productivity with a “pragmatic blend of cloud and data centre”, seems to me to be a reasonably likely forecast. I understand that there is good research emerging to support this position.
A word from the wise
Nothing is new under the sun; Who would have thought that Shakespeare had forecast this moment? About 400 years ago, the Stratford Bard summed up succinctly the counter-intuitive solution that is hidden only because it is so obvious:
“The world is still deceived by ornament…
The seeming truth which cunning times put on
To entrap the wisest”
The Tempest, Act III, scene 2
Do not be deceived by the ornament of Cloud. It’s really good for some things. But that does not make it best for everything.