While there are lots of actions you can take to reduce your AWS spend, many of them require a large amount of analysis and re-architecture of your AWS infrastructure before the changes can be made. There aren’t that many actions you can take that can impact your bottom line today, without having impacts elsewhere in your business.
The advice here is broken down into three categories: easy, medium and hard. The easy actions can be accomplished in a day, with minimal forward planning needed. But many of the harder actions require a detailed, in-depth understanding of how you’re using AWS, and why you’re using what you’re using. Architecture decisions shouldn’t be made lightly, and without buy-in from other stakeholders.
A few of those changes will require many person-hours to make, so any potential savings will be over the longer term.
You can do a lot of this analysis yourself, but Stax’s Cost Management features, especially the Wastage report, make this a lot easier. Stax calculates the dollar values of each of these wastage areas so you know where to focus first. Once you’ve worked that out, the report makes finding the individual resources really easy.
Stax provides you with a list of every resource it believes is wasting money in AWS, as well as the cost of each of those individual items. Why waste your time trying to track down where you’re wasting money? If you’re interested in how easy it is to action Stax Wastage reports, this article explains the process of optimizing costs with Stax.
But whether you choose Stax or not (we obviously have opinions here), here are things you can do that will make a difference.
Today – Focus on the “Easy” Wins
Here are three things you can do today. They shouldn’t break anything or make anyone angry with you, and they will save you dollars.
Delete Unused Resources
Unattached ELBs, Elastic IPs and EBS Volumes can be easily forgotten when you terminate the EC2 instances that are using them. But they cost you money every hour they’re on. Find them—again, Stax can help with this—and delete them. You’ll save money on your AWS bill, and your environment feel much cleaner.
Snapshots in S3
Snapshots are great for backing up system states, data, and lots of other valuable things. You should be taking regular snapshots. But you should also be either deleting old snapshots, or putting them somewhere cheaper if you need them for compliance purposes.
Set up a policy that will move your snapshots to cheaper storage, like Amazon S3 Glacier, after 21 days. If something bad does happen to your infrastructure and you need to restore, it's really unlikely you’ll need to go back to more than about a week before the issue. And if you really do, waiting an hour or so to get that backup out of Glacier will seem like a trivial problem.
Select the Appropriate Storage Class
Check for things in S3 that aren’t accessed frequently and move them to cheaper storage, either S3 Standard-Infrequent Access or Glacier. You can set up lifecycle policies to manage this for you. There’s a good chance nobody will notice the slight delay when they go to use infrequently accessed resources from a cheaper storage tier.
Tomorrow – Research
Now you’ve tidied up the low-hanging fruit, it’s time to do some homework. Look at things that affect your overall cost efficiency, so you can figure out if making further bigger changes will have a negative impact.
Review Legacy Software
If you’ve got a legacy process that requires an old version of an application or database, or one of your software vendors requires a minimum spec, you shouldn’t go changing that without asking the application owner, but you should reach out to the software vendor or the process owner to alert them to the problem. If you can show the software vendor that they’re causing cost inefficiencies, hopefully you can either get a discount on your renewal or you can get them to update their application to be more efficient which will save you money in the long run.
Check for EC2 Instances with Low Utilization.
Do you have instances that are hardly utilized? Consider switching them off or changing their size. If the usage of these instances is on a predictable schedule, look at switching them off when they’re not expected to be in use, by leveraging instance scheduling.
Update Obsolete Instance Types.
Whenever new Instance Types come out, you should adopt them ASAP. The new types are nearly always going to be cheaper than the old ones, so unless you have reasons to keep using the old ones, you should move. Valid reasons include being locked in by a software vendor, having a legacy application that can’t be updated to use the new type, or being locked in by a Reserved Instances contract. Yes, it does take time and effort to move, but by using the cost savings estimate on the Stax Wastage report, you can decide what is worth moving to give you the best return on your investment of time.
Next Week and Beyond
Once you’ve taken care of the short- and medium-term issues, you can look further ahead. These decisions should not be made lightly. There are benefits and drawbacks for every decision, and you’ll need to find a balance between your agility and risk tolerance for these to work well.
Lock in Pricing
Consider investing in a Savings Plan or Reserved Instances. Savings plans are newer and more flexible, but the idea behind both is essentially the same. If you’re using certain things 24/7 to “keep the lights on”, and that amount of EC2 or RDS is predictable and unlikely to change, you should look at locking in a better price.
Depending how you commit and how you pay, you can save up to 70% on your AWS costs, but there’s risk involved if anything changes during the term of the agreement, which is a minimum 12 months. Again, it’s really important to have a good understanding of what your future usage looks like, what your long-term plans are and if there are any projects coming up that will change any usage trends or the savings plan may end up costing you more than you save.
Rearchitect Your Infrastructure
This is where dramatic savings can be made. If you use a lot of EC2, would it be possible to move some of that to AWS Lambda? This can take quite a bit of work to set up, but it tends to be a lot more cost-effective to call Lambda Functions and run code only when you need it, and not have it sitting there waiting for things to happen. That’s how we run Stax, by the way.
But if you do decide you need to use EC2, could you use EC2 Spot Instances instead of On-Demand? Taking time to ask these questions will allow you to plan and make ongoing savings.
Time to Save
Through a combination of all these methods, you can make real savings on your AWS bill. It will take effort and it needs to be done on a regular basis—it’s not a one-off exercise.
But Stax’s cost management tools, especially the Wastage feature, will make it easier by highlighting where to focus your efforts for the greatest return. Why spend hours resizing an instance that will save you $5 this month, when you can spend 10 minutes setting up an archiving policy for old snapshots, that will save you hundreds of dollars over the next 12 months?
Then if re-architecting is a path you want to go down, Stax can help. The platform’s account and Workloads features, in particular, can help you build a more efficient, compliant cloud ecosystem.
Sound good? Get in touch and we can set up a demo.
Topics
