Building complicated architecture on AWS can cost you a considerable amount of money. This fact makes people think twice before moving their infrastructure to AWS. However, AWS offers many ways to reduce costs and a professional architect will know how to optimize your system the way that it would be cost-efficient and suit your business. One of the means that is used for AWS cost optimization is EC2 Spot instances.
Reducing server costs by 90%, is it possible?
Do you often come across services offered at a 90% discount? Well, with AWS as the first and one of the best cloud services providers, this is the reality. You can rent spare servers only for 10%-30% of the usual cost of AWS On-Demand instances. Spot instances are EC2 computing capacity, which is sitting idle at the moment. They are exactly the same servers, so what’s the catch? The difference is that AWS can claim back a Spot instance anytime, giving you a two-minute notice. Despite this inconvenience, knowing how to use them in the system can bring you a significant reduction in your AWS bills.
How we use Spot instances
Spots can be used in fault-tolerant, stateless applications. If you are interested in smooth operation of your system and high uptime, you have to use Spots wisely, applying certain architecture principles, because they are prone to interruption.
EC2 Spot instances are well integrated with various AWS services, which makes them easy to use. For example, Auto Scaling can request them automatically. You can set the rules of using Spot instances in your Auto Scaling group, and the desired proportion of Spots in the group. Auto Scaling will maintain the needed number of servers, if some Spot instances are interrupted, other Spots will be automatically requested to replace them.
As you see, Amazon Spot instances have several advantages, they are:
- Available at a low price
- Well integrated with main AWS services, and therefore are easy to use
- They are the same servers that you would buy on demand
Availability of Spots is not guaranteed, but working with different types of Spot instances and setting other parameters smartly, an AWS architect can achieve the needed level of stability in the system.
Case from our experience: infrastructure cost optimization
One of our clients came to us for hosting cost optimization. They were unhappy with how much money they were paying for their servers on AWS and were going to switch the hosting. We audited the infrastructure and found out that Spot instances weren’t used in the system.
Optimizing this system, we added Spot instances to the Auto Scaling group, replacing 60% of the servers. Spot instances cost was 70% less, which gave us a 42% economy on the Auto Scaling group price, and a 25% economy on the whole infrastructure.
We applied several other cost-optimization techniques, which included resizing of low-utilized instances, applying for saving plans, and others, which resulted in another 18% price reduction. When the client came to us, their monthly AWS payment was around $1500, and after we optimized the infrastructure it became around $850.
Building Spot instances into the infrastructure, the architect should know how to do it without hurting the uptime and smooth operation of the whole system. We have the knowledge and experience needed to work with this type of instances, and use them in our clients’ systems.
If you have any questions, ask us
It is possible that your system can be optimized and you can pay less for your servers. We’ll check it out and give our recommendations.
If you need professionals to move your system to AWS, we’ll be happy to help.