Val, Here is a bit more info, we are building an API gateway and planning on using either Ignite or Hazelcast as our cluster management paltform (it's embedded with Vert.x though). Based on the comparison <http://ignite.apache.org/use-cases/compare/hazelcast.html> and particular Ignite features that we are interested in such as "scaling without cluster restart", "nodes grouping", "dynamic cluster groups definitions based on predicates" Apache Ignite is the winner. At the same time it seems to have much larger memory footprint.
Answering your following up questions: We should be able to handle up to 16 nodes. We expect no more than 5-6 caches in the whole system. The amount of cached data will not be the gigabyte range, probably will be less than 1Gb. The configuration is default, I've also tried to reduce initial cache to very lower numbers, like 10Mb, but it seems not to have any effect on the memory footprint. --Alexander. -- View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Reducing-memory-footprint-tp3494p3518.html Sent from the Apache Ignite Users mailing list archive at Nabble.com.
