At Instaclustr, our vision is to provide organizations the freedom to confidently deliver applications at the highest levels of scale and reliability through open source technologies. Today, we are proud to announce the Instaclustr Certification Framework for Open Source Software and our first certified technology, Instaclustr Certified Apache Cassandra, as the latest important step towards realising this vision.
Over Instaclustr’s 5+ year history of providing a managed service for Apache Cassandra, we have incrementally increased the level of sophistication that we apply to testing Cassandra before we release it to our Managed Platform. Several months ago we realized that we had an opportunity to substantially increase the rigor of this testing, provide open visibility into the process and outputs and provide some real value to our Managed Platform customers, our Enterprise Support customers and the wider Apache Cassandra community. This effort has resulted in the Instaclustr Certification Framework for Open Source Software, a rigorous testing and evaluation program that will apply to each of the open source, data-layer technologies Instaclustr supports.
The Instaclustr Certification Framework formalises the overall process of considering open source for inclusion in the Instaclustr platform. It has two key stages: assessment, where we consider the governance, adoption, and liveliness of the project and certification, where we develop a test plan and undertake independent technical testing of a specific software release or version. Both phases result in a detailed, publicly available report of the work we have undertaken.
For this initial release, we have published the following documents:
- Instaclustr Certification Framework for Open Source Software
- Project Assessment Report for Apache Cassandra
- Technology Testing Plan for Apache Cassandra
- Certification Report for Apache Cassandra
Key highlights of the Certification Report include:
- Performance testing (latency and throughput) comparing the current version to the previous two released version for multiple use cases
- 24-hour soak testing (including repairs and replaces) including garbage collection time
- Shakedown testing against popular drivers
We do view the current testing approach as a starting point and expect to continue to build on it over time. One particular area of focus that didn’t make the cut for the first release will be testing of much larger clusters.
We’re excited to be releasing this work and really hope it will be valuable to the wider community. Please leave a comment below or drop us an email at firstname.lastname@example.org if you have any comments or questions.