What happens behind the scenes in your cloud over Labor Day Weekend? Before you start thinking about lighting up the BBQ, make sure that your AWS cloud is under control and not creating unexpected surprises for Tuesday morning.
AWS offers an amazing pay-as-you-go cloud infrastructure solution. Yet finding the right balance between configuring and maintaining a cloud that is ready to meet your exact business needs can be a challenge in itself.
Since we launched our private beta of 'KnowYourCloud Analytics'Newvem Analytics, a tool that makes AWS analytics easy and cost effective, we've analyzed nearly 900 unique AWS clouds and have seen significant parallels between waste and holiday events.
On Memorial Day weekend, we discovered on average, a 12% increase in the amount of idle instances across our users' clouds, as well as a 10% increase in instances with idle CPU. For heavy users, just one of these small changes can easily result in $10Ks in wasted spend. More importantly, you probably won't find out about, let alone be ready to handle, a potential problem until Tuesday morning.
For this Labor Day Weekend, how can you ensure you won't underspend if demand ramps up? And, conversely, how can you ensure you won't overspend if business slows down over the holiday?
Here are 5 best practices you should consider to help make your Labor Day Weekend 'Cloud-Care-Free".
If you're expecting an increase in demand over the weekend here is a checklist to mark off before you head out on holiday:
* Check for unhealthy load balanced EC2 instances. Unhealthy load balancing with instances are a prime point of failure over AWS. It's not enough to have an instance behind Elastic Load Balancers (ELB); instances need to be configured with a health check to ensure that the instance is working correctly. When instances fail the health check, they are automatically moved from the ELB itself. The risk here is that you could be saving data, running applications, or relying on other loads to execute, but won't necessarily know that there is a problem with data stores until it's too late. Combined with poor backup practices and limited resources over the holiday, you could potentially struggle with long delays redeploying your data, applications or other cloud environments. If in doubt, take a snapshot before the weekend, shut the idle instances down, and reload the snapshot when you return next week.
* Check for over utilization. Over utilization can be an embarrassing, and generally single point of failure over the weekend. In some cases we've seen instances become over utilized because they weren't ready to handle unexpected demand over Memorial Day, resulting in data loss and other problems. Make sure you have or consider an auto-scaling solution to overcome utilization shortcomings.
* Quick Housekeeping. Check for stale snapshots, ensure that your EBS volumes are backed up, check to see if Amazon has any scheduled maintenance set over your cloud during the weekend, and ensure security ports are closed. Otherwise, in the event of an unexpected event, such as hurricane damage or a security breach, your data may be difficult or impossible to recover. We've also seen our customers backup their application environments so they can restore, and ultimately resume business, as fast as possible.
If you're expecting business to slowdown over the holiday or want to shut down part of your application environment, make sure your resource management strategy is aligned with a planned decrease in demand so you don't over-spend.
* Check idle instance count. Idle instances are EC2 instances that you're paying for but aren't using. When gone unchecked, idle instances can run up your AWS bill. If executed over a large batch, or forgotten about over a long period of time, your bill can easily run through the roof. The problem can become more complex when your cloud environment changes frequently and you may not recall what resources are running vs. what you exactly need. Consider an auto-scaling solution to identify and close unused instances over the weekend.
* Check for abnormal changes in cost. Abnormal changes - both increases and decreases in your cloud - can denote substantial changes to your cloud configuration. Cost increases, such as recently purchased resources should be confirmed and aligned with your planned three-day weekend needs. At times, decreases in costs may indicate that you may not have adequate resources to meet planned demand over the weekend.
About Newvem. Newvem is an AWS usage analytics company that makes Amazon Web Services (AWS) easy and effective for cloud users. Dev-Ops, IT Managers, and Business Managers use Newvem every day to discover meaningful insights on their AWS infrastructure while enhancing AWS efficiency and proficiency.
Newvem Analytics analyzes AWS cloud usage, reveals problems related to costs, availability, security, and utilization, and offers solutions to manage clouds more efficiently and effectively. Newvem has raised $4M in Series A funding, with participation from Greylock Partners, Eric Schmidt's Innovation Endeavors, and Index Ventures. Sign up for free here: www.newvem.com