Elasticsearch Capacity Planning Service
Saving costs while ensuring the health and performance of your Elasticsearch infrastructure.
There is no magic formula to make sure an Elasticsearch cluster is exactly the right
size, with the right number of nodes and right type of hardware. The optimal Elasticsearch cluster
is different for every project, depending on data type, data schemas and operations.
There is no one-size-fits-all calculator. Write-heavy workloads require different cluster configurations than read-heavy workloads, and so on. And that’s also why no one can promise accurate numbers and guidance. Until now, that is.
This is how we do it:
We solved the cluster sizing problem with a rigorous, tailor-made, process that ensures the right cluster size and hardware recommendations for the exact requirements of each organization.
Initial discovery phase and data collection
We begin with data request (can be a sample), index mappings, queries, and any KPIs or SLAs you’d want to put forward.
Running initial benchmarks
We run fully automated benchmarks to establish a performance baseline we can then use to create recommendations to support the desired queries and indexing speeds.
Testing on your environment
We begin testing on the exact platform you will be using. We can even use your cloud account. Currently supporting AWS, GCP, Azure, Kubernetes anywhere and virtualized on-prem hardware.
We launch multiple clusters with different configurations as decided by our team. Each benchmark run generates a full report that we compare with previous (and future) benchmarks in order to understand which configuration has the most impact on the KPI being measured.
Discussing & adjusting preliminary findings
After several iterations of benchmarking on various configurations, we will send preliminary findings to confirm our results and proposed direction. This will validate both business requirements and any trade-offs concluded as part of the sizing procedure.
We will present our findings to your team, including results, recommendations and the reasoning behind it, giving you ample time and space to ask technical and business relevant questions.
We are happy to stay in touch and offer support for all your Elasticsearch & Elastic Stack needs and questions.
Our novel sizing procedure guarantees solid conclusions in minimal time, tailor-made for your organization:
Elasticsearch in-depth expertise in all phases of process
Using the expertise of our seasoned Elasticsearch team allows for a multitude of highly effective test configurations between benchmarks, ultimately saving time for fine tuning results and increases accuracy.
Parallel benchmarks for superior accuracy
We run each benchmark in an isolated cluster deployed specifically for that benchmark. This allows us to run many benchmarks fast, which means our recommendations will be based on multiple data points, allowing us to be very accurate.
Business-minded for the relevant trade-offs
Our approach incorporates discussions and validation moments with your technical team and relevant business stakeholders, allowing for business decisions about the necessary trade-offs. This approach prevents any misunderstandings along the way, verifying truly fit-for-purpose cluster size.