Ingest and Retrieve data from an ADX Cluster Us ing Python (2024-05-06)

The DBAccountant

One of the things that I've found with cloud computing services is that the people involved with managing these resources sometimes get asked to become financial accountants.

I saw an interesting post from SQL Rod asking about this new task as something all of us might need to consider a part of our job in the modern world. He asks if we are Techouncants or Accountechs? This isn't in the sense of being the Financial DBA trying to manage costs in the cloud, though that is part of his post. It's more about making smart financial decisions. Certainly, as more workloads move to the cloud, and they are for many of us, we likely need to keep an eye on costs, usage, and tuning.

However, there are other choices. When I had to spec servers, we wanted to get something with room to grow, but not the top-of-the-line most expensive option. I wanted something engineered well for the next few years, and certainly not under-engineered. Usually because adding resources was hard in those days.

Virtualization has made things easier, but often there are still some limits. Still, the goal is matching resources to workload with room to burst if needed. At the same time, are there other choices we can make? Rod asks about choosing the appropriate edition, the HA/DR strategy, tooling, and more. Can we be efficient with our use of money, time, and results, trying to neither overengineer nor under-engineer?

The whole post has a lot of "it depends" in it, as the decisions on what you should choose are something you have to think about and arrive at a decision that balances costs with other factors. It can be clearer often to choose one thing over another, but it's not always crystal clear which choice is better. I like his hardware analogy in the post. I'm not a fan of just throwing hardware at a situation, but in some cases, that's the best solution.

The cloud makes it easier to see where costs are. As a Redgate customer put it recently, "tuning queries becomes more important in the cloud as we can see exactly what each one costs." That can be true for your SQL Server VMs, but it's often not seen as important when the costs are a one-time spend for your hardware. However, Jeff Moden might argue that you should be tuning queries even without chargebacks or a cloud bill. It will save time for your users, which is always valuable. I agree, since I think we have no shortage of work and time is our most valuable resource. Save it when you can for the most people.

That often means a big part of being a DBAccountant is being able to show, lead, or somehow get others to write better code early. Ensure everyone (or automated tests) knows how to view execution plans, learns to use them, and tries to write more efficient code the first time.

Steve Jones - SSC Editor

Join the debate, and respond to today's editorial on the forums