I must admit, I was massively disappointed with the shortage of knowledge on cleaning up crumbs – and I’m talking generic crumbs, not even those specific to pies. If you’ve followed the path of rightsizing correctly and purchased reserved instances to maximize your return with the best coverage, then you’ve probably shaken out your napkin and are ready to leave the table along with the wayward bits of pastry.
Not so fast!
Parts 1 and 2 of this series deal with optimizations that will return chunks of saving. But they’re difficult to repeat often for the same return – it starts to become a game of ‘tweaking’ (not twerking, which is ill-advised after too many pies), and you may think your work is done. Actually, your work has just begun, as the famous Bezo’s pie franchise expects you to clean up after yourself.
So, just like those posh restaurants where the waitron gets stuck in with a tool designed just for the job (if you think I’m kidding, ask Eddie what he’s using), here are a few ways to scoop up the wayward bits.
Unattached EBS volumes.
Because it’s scary to delete these, or because you opted not to delete root volumes when instances are terminated, EBS volumes can float around your infrastructure for years. You’re paying full price for these, so find them and delete them (snapshot before-hand for peace-of-mind) and save yourself a few points on your costs.
There’s a catchphrase from post WWII that goes “Old soldiers never die, they simply fade away”. Well, old EBS snapshots never die either. Nor do they fade away. And they may go back WWII (in internet years at least). Despite the recently released Amazon Data Lifecycle Manager giving you the ability to employ retention and deletion policies, I’ll take a bet most people still have very old snapshots, too many snapshots per volume, or snapshots for volumes that are long gone. Even Jeff Barr will tell you “It turns out that many of our customers have invested in tools to automate the creation of snapshots, but have skimped on the retention and deletion”. Clean it up, save a buck.
Disassociated IP addresses
Elastic IP addresses are a great tool to keep a public IP address available for mapping to EC2 instances in need, and you tend to hang onto them for that reason. However, they cost you more when they’re not associated with an instance so it’s important to check how many you have that you may have forgotten about. It’s not going to buy you a pie franchise, but it will save you a buck or two.
Newer instances are often (almost always?) cheaper than previous generations. That means more bang for your buck, and fewer bucks for your bang. That’s probably confusing, so take a look at this comparison of pricing for the previous and current generation of two instance types (taken from www.ec2instances.info for the Ireland region).
More power (usually due to newer underlying hardware) and less expensive. It’s worth the migration.
Keep your waitron on call.
Here’s the thing with crumbs: They reappear without warning, like reality TV stars, and often where you least expect them. This is where proactive, centralized governance becomes crucial. For example, you could create a workflow to notify admins of unattached volumes or old snapshots, with a deletion option (automated or through a workflow) if defined thresholds are exceeded and the tags match your criteria. Send a notification if instances are running on old generations, terminate ‘Zombie’ instances (after automatically snapshotting)…there are many ways to use best practice policies for cost reduction and optimization to make sure you are getting the most out of your environment.
However, in our experience, customers seldom find the time and resources to implement this effectively. Even those who make a great start tend to lapse in diligence for periods of time because other projects take precedence, people leave the organization, or it just isn’t anyone’s day job.
That last point is perhaps the most crucial, and it’s why we have tailored our cost optimisation and management service to provide deep insight into initial savings opportunities and to become your cloud steward (or sommelier if you’d prefer that) for ongoing optimisation and governance. Our chosen management platform, CloudHealth, extends across clouds and beyond optimization to categories including security, financial best practice and performance too if required, to ensure you manage your cloud rather than have your cloud manage you.
It’s our day job, and we’d be thrilled to run a free assessment for you.
Russell Warne is Chief Customer Officer for Kaskade.cloud. He is a Certified AWS Solutions Architect – Associate and a Certified Cloud Health Platform Administrator Associate for Cost Optimisation.