| Seems to me that this sort of thinking is predicated on the the idea that infrastructure should still work like it did 10-20 years ago. Cost is not the driving factor, and in any case, cost is often calculated wrong. Putting one set of prices 'for stuff' in one column and another set in another column and looking at the difference tells you almost nothing. Forget about start-ups for a minute. There are certainly arguments to be made both ways there. Focus on big enterprises for a second. Infrastructure complexity isnt getting smaller over time and neither is demand. Delivering on demand and managing complexity with a bounded number of people requires a change in thinking when it comes to infrastucture writ-large. We cannot sustain the old on-premises, dinosaur pen style data centres and deliver and grow our core businesses. It just wont work... So you either go cloud or create something very cloud like and do it on-premises. Heck that is how AWS came about in the first place. Anyone that thinks going cloud is a good way to reduce headcount is going to get a shock. Going to cloud (or lets say changing the way you do infrastructure) is a way to continue to do business with the headcount you have. It's not a question of carrying on with the 'old school' infrastructure you have, you just want keep up doing that. Do nothing and your headcout requirements are unbounded. Anyone that thinks cloud infrastructure requires you to hire only a bunch of cloud experts is wrong. Chances are you have the bulk of the infra people you need right now. All those people that have been doing 'old school' infrastructure for years are still your most valuable resource. The mechanics of the infrastructure are fairly irrelevant (ok they are not, but in the grand scheme of things we can kind of cancel the mechanics out in the equations), oft missed value is the operational knowledge. The operational bit is what gets swept under the rug in the DevOps discussion. I believe the knowledge of how to translate existing infra into a 'new' model doesnt come from hiring DevOps people or cloud people, it comes from the infrastructure people that have been doing it for years. Leverage the intellectual capital you already have. The flip side of that bargain is that 'old school' infrastructure people need to recognise we have to adapt. Those that dont are doomed to be cancelled out in the same equations that cancel out the mechnanics of the infrasturcture itself. Enterprises that fail to recognise the shift are also doomed. Those start-ups we arent talking about? They can scale far more quickly and quickly get to a point where they can deliver at the level a much bigger organisation can. They can eclipse the slow movers. I assume this is why a lot of consolidation happens... The slow dudes only really have one move and that is to buy the little guys before they can get there (the Facebook defence). But thats like doubling your bet on red every time you lose at roulette. Eventually you go bust or the game moves quicker than your bankroll... Thanks,
James |