The Pros and Cons of Amazon OpenSearch Service

Cons of Amazon OpenSearch Service

Limited Control. You are limited to the parameters set by AWS.  For instance, you can use a maximum of 40 data nodes per cluster.  However, some clusters require 60 nodes or more. 

You will also be limited to whatever specific version(s) of OpenSearch that Amazon makes available. And importantly, be aware that at times Amazon may require you to update OpenSearch, even if it’s not in the best interest of your infrastructure. For instance, a forced update can cause outages and breaks if other components of your infrastructure aren’t compatible with the new OpenSearch version. 

Costly If Not Diligently Managed.  While Amazon OpenSearch Service can be cheaper to initially set up than working with a managed services company, costs can quickly get out of control. If your organization is not diligent in monitoring and optimizing resource usage, you can end up with surprise bills where costs far exceed the initial budget. 

These large costs are partly because the AWS pricing model can be complex (perhaps purposely so). And further, because companies are removed from the setup and installation process, Amazon is incentivized to oversell hardware.

Working with an independent OpenSearch managed service provider, like Dattell, mitigates this problem. An independent provider does not benefit from you over-paying for cloud services. The provider also understands pricing and has experience with lowering cloud costs without compromising system performance.

For our larger clients with expansive OpenSearch infrastructure, we’ve saved many of them over $1M per year on server costs. This is a savings well beyond the cost of our services.

And while many of our clients are large corporations, we also work with small and medium sized businesses. We’ve seen this overpaying for cloud services even for smaller clusters.  For instance, with a client that had only 40 employees, we saved them enough on cloud costs to hire an additional employee. 

A few additional thoughts on the unexpected extra costs when using Amazon OpenSearch Service:

• While users can add new servers at anytime, you are limited to whatever is available.  Therefore, you may need to pay for a larger server than you need.
• Because it is easy for employees to add new servers, they aren’t incentivized to code or otherwise optimize infrastructure.  This adds unnecessary bloat and cost to the system.
• And finally, running OpenSearch in any cloud will be more expensive than using on-prem servers, especially for performance projects such as machine learning. 

Service Outages.  Like any cloud service, Amazon is not immune to occasional outages. Additionally, Amazon can and has brought down servers for maintenance outside of the provided schedule.  We have seen from firsthand experience how this creates negative impacts for companies using Amazon’s service.

Learning Curve.  Amazon OpenSearch Service requires a learning curve. Understanding the nuances of indexing, querying, and managing the cluster effectively can be challenging.  Consider who will train your team.

Vendor Lock-In.  Opting for Amazon OpenSearch Service means you’re locking yourself into the AWS ecosystem. While AWS is currently a trusted provider, it’s essential to consider the long-term implications of vendor lock-in. 

Migrating away from AWS can be complex and costly. Consider, will locking your infrastructure into Amazon align with your organization’s long-term strategy?

Learn about managed OpenSearch on Azure.

Limited Support.  Consider how long it will take to get assistance for preventative maintenance and more importantly, during a production issue.  Will you be working with a support person who knows your use case and needs, or a random on-call engineer who will need time to get acquainted with your infrastructure?

Also, consider if the amount of optimization, preventative maintenance, and training provided meets your current and future needs.