Are You Embracing a Holistic AWS Cost Optimization Strategy?
- AWS & Azure
- Jordan Jacobs
Amazon continues to cut the prices of its cloud computing services — recently reducing the number of Simple Storage Service (S3) tiers from six to three, resulting in a 16 to 25 percent price reduction. Meanwhile, AWS Glacier service for storing long-term, infrequently accessed data received a 43 percent price drop.
It’s all part of the “race to zero,” wherein major providers slash costs to encourage long-term adoption.
But there’s a catch: Falling prices do not inherently lead to smaller spends. Many companies still struggle to reduce public cloud overhead, maximize resource efficiency and prevent resource sprawl.
Companies need to implement a holistic AWS cost optimization strategy. And very few that I talk to are doing it. The following will get you a head start.
The Basics of AWS Cost Optimization
First up, make sure you’re following the AWS cost-control basics. This includes cleaning up orphaned snapshots after you delete an EC2 instance — while the EBS volumes are automatically deleted, many companies leave the snapshots behind, which in turn consume valuable resources to host and maintain. It’s also a good idea to tag everything. You can create up to 10 tags in the AWS console. These key/value pairs let you monitor and track usage across EC2 instances, S2 buckets and EBS volumes — and help you identify areas where you may be overspending or underprovisioning.
It’s worth mentioning data transfer costs, too. Most companies know that sending data into the Elastic Compute Cloud is free, but moving it out is not.
If you’re moving data between AWS regions or transferring out of the Amazon cloud altogether, it’s easy to blow your budget and overspend to ensure all your data makes the trip in a reasonable time frame. While this may not be a cost you can avoid, it pays to anticipate this expense upfront.
Rightsizing for Your AWS Applications
Now that you’ve got the most popular cost reduction strategies covered, it’s worth mentioning the other big one you’ve probably heard about: Rightsizing. Amazon offers this solution for EC2 via managed services, which provides an analysis of your current cloud needs and then recommends the ideal mix of solutions and resources to maximize performance while reducing costs.
Understandably, some IT managers are worried about getting cost-reduction advice from the company they’re paying for cloud services.
SingleHop offers an expert look at your current and potential cloud needs to determine the optimal blend of EC2 and RDS instances for individual applications. First, instances are monitored to give you data on CPU usage, memory usage and total performance, making it easy to categorize and prioritize specific instances. Next, recommendations are made to shift away from the typical 30 percent CPU utilization common among most cloud customers. Once implemented, you should run closer to 80 percent — enough to make sure you’re getting your money’s worth with enough room to handle any sudden resource demands.
The idea behind rightsizing is simple: Instead of spending on more than you need simply because AWS prices are coming down, you maximize performance and budget by running applications in the ideal environment.
The ‘Get-Out’ Gang
The biggest component missing from most AWS cost-control strategies may be the most obvious: Sometimes, you’ve got to get out.
Let’s start with Reserved Instances (RIs). These fixed-capacity cloud solutions are available at a steep discount from Amazon — you can save between 30 and 75 percent compared to on-demand solutions — and they provide a “capacity reservation” in a specific availability zone.
Seems ideal, right? You save money and carve out the space you need for particular apps. Not so fast. There’s a caveat here, and that’s the need to pay your RI term in full and upfront. Terms are either one or three years, and there’s no money back if you don’t use any portion of RI resources during the term. The result? It’s easy to purchase an RI, discover it’s not enough (or too much) for your app and then stop using the instance, meaning you lose money and time.
In short, it might be best to get out of the RI game unless you’ve got a consistent, predictable workload that won’t change over the next 12 months.
The other side of this often-overlooked cost-cutting measure is to take the hybrid cloud approach.
Think about it like this: Once you’ve started the move to public cloud computing, especially AWS, it’s tempting to go all-out and move every application and service, especially with cloud security now on par with local protection and on-demand scalability limiting the possibility of a resource crunch.
But here’s the thing — some apps simply don’t belong in the public cloud.
Maybe they’re holdovers from legacy development, handle mission-critical corporate files or are required for compliance and regulation audits. Here, it’s a good idea to get off public servers and instead opt for a hybrid solution: Run some apps in EC2 and others in a non-AWS dedicated environment. Your savings? You don’t overspend on cloud resources you don’t need, or watch your costs rise as sudden traffic spikes force you into pricey on-demand territory.
AWS costs are coming down, but that doesn’t mean the public cloud is always your most cost-effective choice. Leverage the basics, right size your workloads and get out off RIs or into hybrid cloud as required to reduce your total spend.
Explore Managed AWS Cloud With a Solutions StrategistSchedule a consultation
Leave a Comment