Spark and memory usage

Spark as a business IM product is fantastic–feature-wise. It has important features that work well for a business environment, but at the same time it keeps things simple and to the point. I like the product from a feature standpoint. What doesn’‘t work for our environment, though, is its memory usage. I know this all stems from the fact that it runs on the Java platform. Each user has their own virtual machine userspace when Spark is running, and therefore it’'s easy to see 30-40MB of memory usage.

Why is this such a problem? It just absolutely is not feasible in a terminal server environment. It severely restricts the number of users who can be packed onto a server vs using other IM solutions for the client. My question is: are there any plans in the future (even if it’‘s distant) to move to a different platform than Java? If not, are there tweaks when using Java to keep the memory usage down per instance of the client so that multiple users don’'t add up to hundreds upon hundreds of MB of memory usage?

Like I said, great product. I’‘m just saddened that optimal memory usage wasn’‘t a concern in it’'s plans. It really does make it non-practical solution for such an environment as a terminal service one.


it may be possible to decrease the memory footprint again about 5-10 MB but not to 5-10 MB which would be fine for a TS environment. Maybe a browser-based client like JWChat would be there much better.


We have also gone over to LinQ because of the huge amount of memory that Spark use.

Look at this reply: