VesiroSearch is now in beta, which means that teams that run Elasticsearch can download and evaluate it using their own data and queries. The beta is free for the duration of our beta period for up to five nodes.
It installs as a standard plugin and the installation process takes about a minute. It doesn’t change the query DSL in Elasticsearch, which means that you do not have to change your queries or mappings to run it. No other end points are affected either, and VesiroSearch works with your current indices, so your setup remains the same.
VesiroSearch is a purpose-built C++ search engine that intercepts the query phase and executes it instead of the standard Elasticsearch.
The exact increase in performance depends on many variables such as query complexity, document size, number of documents, memory to index size ratio, etc. In customer pilots, we’ve consistently measured up to 2x faster search on indexes of up to several terabytes in size.
The improvement in search performance means that you need less CPU resources to handle the same search workload per given time unit. That enables you to do a few different things. If you change nothing you will be able to get much higher query throughput with your current server infrastructure. If you are looking to reduce costs, you can choose to reduce the number of server nodes until you get to the same level of query throughput you had before. You can also choose to move to cheaper server nodes.
This is mainly for teams running self-managed Elasticsearch clusters (on-prem or cloud doesn’t matter) at such a scale that cost has become an issue. The increase in search performance is useful if compute is part of your bottleneck. You can request access to run the beta version of the product. It’s free for the duration of our beta period for up to 5 nodes in total.
We would love to get feedback from teams that run Elasticsearch extensively. Learning which workloads that benefit from our acceleration, and which workloads we should add in our product roadmap would be super helpful. Also, if you have any feedback on compatibility or what would make evaluation easier, that would be greatly appreciated as well.
You can read more about the product here: https://vesiro.com And you can request access to the beta here: https://account.vesiro.com