My company is currently running Spark 2.5.8 as a pilot. It runs as needed and we like it but we are being asked to scale it up to support 10,000 clients. Does anyone have any documented experiences for Spark scalability at the server level;either Windows or Linux? Need to know how to build the server platofrm out to support the larger client base but need to know if 10,000 is realistic given the single server architecture.
Spark isn’t the issue. For scalability you will have to determine if your XMPP server can handle the load. You didn’t mention what server you are actually using or whether you expect to support 10,000 users concurrently, or what features those users would be using. These are the factors that will affect scalability, not your client.
If you’re running openfire in linux then you need to increase you max open file handles to the number of lcients you want connected plus about a 1000.