AIOps customers are some of the largest in the world with the largest amounts of data.  In 4.9.1 we are now ready to support a steady event rate of 1,000 events per second. That is 86.4 million events per day! A 43% increase from the previous release. In addition we have increased support for topological resources 33% to 20 million resources, doubled topological groups to 60,000 and doubled our standing alert count to 400,000. These scale points can be achieved via custom sizing.

As a proof point, we already have one customer as an early adopter who has 400,000 standing alerts in their system today.

Increasing scale from an already large starting point involved a lot of effort from multiple teams! Being able to process, store, query, and present so much more information meant we had to optimise our UI, our backend datastore and the storage we use.

In the backend, we distributed our data and now have it available from multiple services to allow more concurrency. This means we now have more services to distribute the load, and more resources to serve them. We also optimised our underlying datastore by working closely with our partner. We analysed and optimised our queries, and we reduced the frequency and quantity of data we send to the UI.

Further we have closely investigated the storage we use and found that some of our technology with high disk I/O performs better and is more reliable on local storage, rather than using a storage provider.  There are a few reasons for this, but perhaps the simple explanation is that distributed, network storage providers replicate data on top of our data replication at the application layer. Removing this additional layer of replication overhead has improved performance without sacrificing resiliency. We also have a benchmark tool so that customers can better determine if their storage is sufficient and performing as well as expected. With these improvements, our most critical, I/O intensive data systems can read and write to disk rapidly, keeping pace with larger workloads without sacrificing resiliency.

Now when you install AIOps 4.9.1 you will be able to unify more data than ever with faster UI response times!