In this episode we’re talking about all things sizing - clusters, sharing, indexing and we’re joined by Jan Srniček of Global Logic.
We first came across Jan via his talk from MongoDB world, where he spoke about the journey he and his team took in understanding how to reduce usage & cost - all the while keeping performance and responsiveness high.
In this episode, Jan talks about his journey with MongoDB, moving from Cosmo DB to MongoDB. Initially they stood up their own on-prem MongoDB database to learn more, but then soon realised that moving to Atlas was key, particularly as they could host on Azure.
For Global Logic's client, Catalina, Jan manages everything from small clusters with a few hundred records all the way up to a system of 7 clusters with over 5Billion data records!So he know’s all about scaling - and teaches us some lessons he’s learnt along the way. He illustrates that if you only scale one large cluster (e.g with Autoscaling on), your database never gets a break! However, if you have smaller clusters, all autoscaling, along with predictable traffic patterns visualised by using Atlas metrics UI and performance advisor, you can analyse usage and re-organise structure appropriately.
Ultimately, understanding the workloads in your application and dividing those across clusters all working together is key to application performance.
Ján Srniček - https://www.linkedin.com/in/j%C3%A1n-srni%C4%8Dek-4a3b826b
Global Logic - https://www.globallogic.com/
Catalina - https://www.catalina.com/
MongoDB - https://www.mongodb.com/