There are all kinds of reasons why companies undertake cloud migrations, but it’s probably fair to say that many of them are likely to revolve around the potential to reduce costs. Cloud environments can indeed offer the potential for significant cost savings but in practice they are only going to do so if you manage them properly. In other words, if you’re prepared to get to grips with cloud cost monitoring and optimization.
Cloud Cost Monitoring and Management
Tools can be helpful, but only if you actually understand how to use them effectively
Amazon Web Services, Google Cloud and Microsoft Azure all have tools you can use to monitor and optimize your cloud spend. There is also a wide variety of third-party tools for the same purpose. These range from offerings which are intended to be “one tool to rule them all” to tools which are intended for one, very specific, aspect of cloud cost monitoring and optimization, such as the performance of your applications, the health of your database or the standard of your network data-flow.
If you’re a small SMB, using just one public cloud vendor, these are very likely to be more than you need. If, however, you’re a larger SMB, they could be very helpful even if you’re just using one public cloud vendor and especially if you’re using multiple public cloud vendors. Tools, however, are generally only as capable as the person using them, so if you want to make any cloud cost monitoring and optimization tool work for you, you’ll need to understand the five basic principles of cloud cost monitoring and optimization.
1. Your cloud strategy needs to manage your cloud infrastructure actively
When you do your initial cloud migration, you’ll have a clear map of your entire cloud infrastructure. You might even have taken your cloud migration as an opportunity to clean up your systems and make them more streamlined and efficient.
The problem, however, is that the cloud environment makes it just so easy to add a virtual machine here and some storage there and create a dependency somewhere else. To be fair, it also makes it easy to remove them, but humans don’t necessarily remember to tidy up everything after their work is done, especially not when they want to go home for the day (or weekend).
This means that your cloud strategy needs to include a process whereby you actively and regularly check your cloud infrastructure so you know what you actually have (or had and terminated) rather than what you think you have (or had and terminated).
2. You need to understand what you do to resource it effectively
Similarly, you need to understand what the people in your organization are doing so you can provide them with exactly the level of resource they need, no less and certainly no more. Remember that one of the major benefits of cloud computing is that it gives you the opportunity to finesse resources to an impressive level of detail.
Probably the most obvious example of this is the way you can adjust the speed of your storage to suit the application. It’s obvious that you’d use your fastest storage for the most business-critical applications and your slowest storage for tasks like data warehousing, but look out for opportunities to rein in the speed of all the tasks in between. Just turning down the storage speed a little bit may be a very minor inconvenience to users (in fact they may not even notice it), but it could save your organization a lot of money.
3. Monitoring cloud cost monitoring and optimization needs to be a team effort
As a bare minimum, there needs to be two people with the skills and knowledge to undertake cloud cost monitoring and optimization in an effective manner. That way, if one is off, for any reason, or leave the company, the other one can cover. Two, however, is cutting it fine and will probably only be feasible in the smallest of SMBs. Ideally you want a team of people sharing the work in an agreed manner (so as to avoid duplication/replication of responsibilities) and each team member should have their own cover.
4. Access control matters, a lot
Basically, you want to do everything you reasonably can to limit the number of people who can buy reserved instances and you absolutely must do everything you possibly can to limit the number of people who can authorize on-demand usage. In simple terms, you want to create a situation where every cost is owned by someone who is held accountable for it. Ideally, you want this accountability to be incorporated into their job description and performance objectives.
5. Use automation as much as possible
Automation is your safeguard against human forgetfulness. If you really don’t want to take the risk of automated tools closing down resources without human oversight, then have them send alerts and make sure humans act on them.