Generally, as with any performance-related question, you'll get a "it depends" answer. Every Splunk environment is different and every customer has different requirements, use cases, usage patterns and so on. So the general guidelines outlined in sizing guide and taught in architecting class are just that - guidelines for a typical scenario. Then the details creep in. The number of data sources or "kinds" of data sources shouldn't _directly_ affect the number of indexers. But if you have certain limitations regarding those data sources and maybe additional constraints regarding performance for different kinds of data it might affect your overall architecture up to the point of spinning up a completely separate cluster for specific type of data. It's all in the details. Oh, and retention on its own doesn't affect the _number_ of indexers that much either. Again - directly. If you have additional physical constraints because you simply can't get more than X TB per single server, that might indeed force you to scale your environment horizontally just to fit all this data but it's not because of Splunk requirements but more because of the hardware limitations. And then comes the smartstore...
... View more