Coherence Topology Proposal

Data for caching:

  • 100 GB data
  • 500-5000 byte objects
  • 1000 objects updated / inserted on average per minute (peak 5000)

A proposal for a coherence topology in production and testing is required (distributed with backup)

  • number of servers
  • nodes to the server
  • heap size per node

Questions

  • How much free available memory is required per node compared to the memory used by cached data (assuming 100% usage is not possible)
  • How much overhead does 1-2 extra indexes per cache element generate?

We do not know how many read operations will be performed, the solution will be used by clients, where the critical value of low response time (more than data consistency) depends on each use case. The cache will be updated from the database by polling at a fixed frequency and filling the cache (since the cache is a data master, not a system that uses the cache).

+3
source share
2 answers

The rule of thumb for determining the JVM for coherence is that the data is 1/3 of the heap assuming 1 backup: 1/3 for the cache data, 1/3 for the backup, and 1/3 for the index and overhead.

, . .

JDK 1.6 JVM 4 , 75 -. (16 ), . (, 16 ) , 1/3 1/3 . , 16 , .

1 ​​ node.

, ( ) . , , 32 , , ? ( ), , . , 75 .

, Coherence 250 ( ) 1K, . , , , . / , .

+5

" ":

1) .

2) Java 7 (, 27 ) G1.

3) 100 , , 300 . 128 3- , 4 JVM 27- (~ 324 ).

4) . , , , , .

0

All Articles