At Rackspace I have a number of business customers who are committed to a Microsoft architectural stack. Some of these are older businesses, not startups born with cloud tech choices nor with access to cloud-era talent. Their stacks include SQL Server exclusively, and it has been part of their world forever. Recently as I’ve been talking with their IT leadership I’ve been particularly curious to ask about their work around cost architecture, and this often leads to asking if their view of workload on their database.
I thought it’d be interesting to reflect on these traditional businesses, and some of their challenges.
For example, there’s a common situation where say a third of a customer’s infrastructure compute is consumed by SQL, with a traditional architecture where the application assumes high-availability from the layers beneath it. Their app has little message queuing or service-orientation. The consequence of their architecture is a requirement for very solid infrastructure.
I’ve been talking to enterprise IT folk about the complications and process of migrating services to the cloud for about four years. Whilst all of us want to see service-oriented IT more widely-deployed, the reality is that this modernisation takes time.
So whilst a business’ IT might be a while away from being a distributed system, it seems sensible to at look at the compute offload, cost reduction and scalability offered by NewSQL.
Migrating to NuoDB or Clustrix can be a much simpler proposition than refactoring to use NoSQL. Existing MS SQL Server databases can be ported, with complexity depending on the usual factors (extensions, size of code base, coupling). It is worth a proof of concept to assess the change effort and benefits.