I recently spoke with Rachel Chalmers, Research Vice President - Infrastructure Management at 451 Research, about being cost-conscious while managing Amazon Web Services (AWS).
Chalmers said you can save a ton of money by introducing cloud resources, and then you can spend it by having to rebuild or go back to fix mistakes.
There is a lot of anecdotal wisdom about cost consciousness coming from usually large enterprises in this realm because they have very complex interdependencies, compliance, and regulatory needs. They oftentimes have very arcane applications that need unusual kinds of support.
One of the really beneficial effects that Amazon has had on the market is providing a cost model for varied commodity services. It has forced these very large enterprises to do some introspection and come up with cost models for their very unusual, custom sets of applications.
Once you’ve started building a cost model around those, you can also start building a model around support and a model around migration. Chalmers says this has brought a lot of discipline into the market that wouldn’t have happened if AWS hadn’t appeared as a competitor.
You need to make choices around performance and health and you need to understand what your needs are and what you’re prepared to pay to meet those needs. You can design for failure until you’re blue in the face, but you really need to know how much up time you actually want. There’s not much point in moving to the cloud unless you’re going to take advantage of the elasticity and parallelism that it makes possible, and when building applications that take advantage of very large data volumes, be intelligent in data placement.
This means you should keep data that changes a lot as close as you can to the compute, and data that doesn’t change very much as close as possible to the cheapest storage tier.