The little accountability paradox that’s costing your company big bucks...
Is it better to be blamed for not turning off an unused AWS server than be responsible for switching off the server that hosts your company's top client configuration? Depends on who you’re asking. Why would a new-hire developer take that risk? And yes, those are the kind of stakes that may run through their mind when faced with a unfamiliar set up and hierarchy: Best to just leave it to someone else.
This scenario is what a developer-lead recently dubbed the “fear factor”, he explains:
"One thing I've seen contribute a lot to AWS costs is a 'fear' mindset thing: as teams change, often the developer tasked with replacing something is a new hire, and since he isn't familiar with the stack it's 'safer' to leave instances running than to switch them off, even though they almost certainly aren't doing anything anymore."
Fire & Forget
As this article from Datamation points out, one of the major costs of cloud computing isn’t actually based on using the service, it more related to organization or lack of, they called it:“Fire and forget: Spinning up a bunch of virtual machines and then forgetting to shut them down when you are done. If they keep running, the meter keeps ticking…”
While it may be a question of simply “forgetting”, we reckon it’s a lot to do with the fact it’s not clearly defined who is responsible for what. One of the “special features” of the cloud-infrastructure environment is a lack of control andaccountability when it comes to attaching a dollar value to choices made (or not made as the case may be).
The fact is, in any other job, employees are ideally provided with the minimal resources to do their job. You don’t see a store clerk making big orders from suppliers, and if you to the quality of management, could be questioned. There is usually a procurement process, to align responsibility with hierarchy. And not just for the sake of the budget, so that everyone can do what they do best without the stress of making budget calls outside of their skill set. But up till now for cloud infrastructure there just hasn’t been. As discussed in this article.
Added to this hazy cost responsibility is that fact that cloud-infrastructure providers haven’t been active in remedying the situation, this 2009 forum post suggests to AWS:
“a little algorithm that says something like = if instance not used for more than one day out of last 7 OR new instance created 3 days ago and not used for last 2, send email to user saying instance is still running.”
It seems they haven’t had time to get around to implementing that in a user-friendly way quite yet, and why would they: fire and forgets are boosting their bottom line. (OK, admittedly AWS have since set up the Trusted Advisor support site, but that’s just some recommendations, it won’t actually stop redundant AWS servers running).
Wouldn’t it be easier to just take this kind of room for error and senseless overspend out of the equation? We thought so. That’s why we developed a workflow that automatically stops servers that are left running for obsolete projects, and sends alerts to the right people about those costing more than they should. We figured it’s time totake control of cloudinfrastructure spend.
Can you relate? Who's responsibility is it anyway? Why is our industry so vunerable to the accounability glitch? Leave your thoughts below and we'll get one of our expert team to respond.