Complex Factories

I spoke earlier about the perception the world has about our world of IT – it’s just one big confusing “technology” mess that is impossible to understand. (Am I being too extreme?)

But there is an enormous divide and, from the outside, it is quite confusing. The split is between infrastructure and business application. I'll talk about this split and it's future later.

However, infrastructure consists of the building blocks of any corporate IT environment – PC’s, servers, storage, LAN’s, WAN’s, firewalls etc. These components and their management should be viewed as a commodity based service that should always be undergoing cost cutting initiatives.

In the world of infrastructure the game is generally to “keep the factory running” so a quest for resilience and stability should be the primary drivers.

In most large IT departments this is done by what most outsiders would consider to be an anal approach to project work. Applying rudimentary engineering principles to any changes like proof of concepts, prototyping, detailed research and up front design with careful and rapid roll-back planning and, unfortunately, migrations that are carried out over weekends or holiday periods.

The reason for this level of diligence is that getting an infrastructure project wrong generally means if it fails nobody can do anything and I mean anything. Ever tried logging on when AD has become corrupted or the core switch is down?

These disciplines are generally at odds with the project sponsors who see no need for this level of diligence. “After all it should just work eh?” At least that’s what the salesmen told them.

The problems are always caused by the complexity of the data that is held in the components of the infrastructure and how hard it is to migrate it from one platform to another. It’s pretty easy to understand what I’m talking about if you’re moving data from one disk system to another but the problems can be just as complex moving firewall rules or routing tables across to a new box.

This migration complexity is a massive hand break on any opportunity an organization has to upgrade or shift platforms and can lead to product entrenchment way beyond any reasonable expectations.

I wonder how many organizations made a decision on their new $5 million SAN 7 years ago based primarily on a juvenile analysis of cost/byte and bytes per second and are now wondering why they have to fork out another $5 million to move to a new one today.

When considering investment in core infrastructure it’s important to understand what it will take to get into it but more important is to consider what it will take to get out of it in the future. Planning for the long term is generally at odds with the pressures of our industry but when it comes to ”factory design” it is of up most importance!