Wednesday, December 07, 2011

Future costs and Cloud

There are many subjects which I find tiresome but two which are starting to irritate me are the notion of Enterprise Cloud and Financial ERP. I'll deal with Enterprise Cloud in this post.

The shift from products to utility services inevitably incurs various forms of risks. These include disruption risks such as loss previous skillsets and political capital to transitional risks such as changes to governance and transparency of suppliers to outsourcing risks such as pricing competition and loss of strategic control.

A common, past method of dealing with transitional risks is the use of a hybrid model combining both public and private supplies. However, this is a transitional approach and should be undertaken with a view of moving to a future hybrid model of multiple public providers (i.e. a competitive market).

A transitional approach requires you to build in a way which is likely to be compatible with a future public market. For infrastructure this mean use of commodity components and in most scenarios an EC2 / S3 / EBS like interface. Hence my general support for open efforts like OpenStack (and to a lesser extent Eucalyptus).

Unfortunately, many applications are designed with the best practice for a product world i.e. scaling is about bigger machines, resilience is about N+1 and in general the focus is on reliable hardware. Best practice for a utility world involves resilience, scaling and failure modes built around software i.e. design for failure, distributed systems and chaos engines such as Netflix's chaos monkey approach. There is an inevitable cost of architectural transition from one set of best practices to another.

Obviously many companies don't like this architectural cost of change and hence want to minimise it which has given rise to the concept of the Enterprise cloud i.e. it's like cloud but without the commodity bit.

It should be noted that cloud is simply a result of a standard process of evolution that inevitably leads to operational efficiency through provision of a commodity. A consequence of this is it also enables higher rates of innovation for new business activities (such as big data) through the combined effects of componentisation and creative destruction. The upshot of this, is that you've never had a choice with cloud - it's just a question of when and the longer you leave it then the more you put yourself at a competitive disadvantage to others.

Whilst a commodity based private cloud (which can and should achieve much lower costs than public provision today) is a viable option in the short to mid term (depending upon scale), unfortunately Enterprise clouds don't move you along that architectural transition and here there's a real gotch'a. The problem is simply known as Jevons' paradox.

As competitors gain the benefits of more efficient commodity provision and higher rates of creation, this is unlikely to result in a reduction in IT budgets but instead more IT activities undertaken with everyone trying to keep up with each other. We've seen this for the last thirty years i.e. as IT has become more efficient, IT budgets haven't fallen but we've just ended up doing more stuff.

You therefore have to factor in that five to six years from now your architectural transition costs may well have spiralled by an order of magnitude due simply to the increased size of the estate. This can obviously be counter balanced with a simplification strategy, assuming your estate is already bloated but when considering Enterprise cloud, you must include this increase of architectural transition costs along with less efficient provision during that time due to a non commodity approach. Even private clouds are going to look dubious in this timeframe.

In most cases, Enterprise cloud will be a pretty unattractive option with ongoing and increasing costs.  It can however still be useful as part of a sweat and dump strategy for legacy environments i.e. you push the capital costs for legacy onto a provider with a view of dumping that part of the estate in the near term.

Why do I find this subject irksome? I simply hate repeating old ground and this has been covered many times before over many years. This is the last time.

-- 26th Nov 2013