The array/stripes of sub maps underneath each CDM/CDSM I think serves two purposes. The first is that any one clustered object lives exclusively on a single active terracotta server. Without the submaps the data would homed on a single server and you may not take advantage of a cluster with multiple active servers.
The other reason I think has to do with getting a more mutation concurrency. At the lowest level in the terracotta server a clustered object can only have changes applied to it by a single thread. -----Original Message----- From: tc-dev-boun...@lists.terracotta.org [mailto:tc-dev-boun...@lists.terracotta.org] On Behalf Of Sergio Bossa Sent: Thursday, April 07, 2011 2:59 AM To: tc-dev Subject: [tc-dev] Purpose of concurrency level in ConcurrentDistributed(Server)Map Hi guys, I was looking at the ConcurrentDistributed(Server)Map code and I was wondering what's the purpose of having the concurrency level setting and so use multiple ConcurrentDistributed(Server)MapDso objects. I mean, Terracotta already implements (distributed) lock striping with lock strategies inside *MapDso objects, so why having another (higher) level of striping/indirection? Maybe I'm missing something obvious, but in the end I'd like to understand how to properly configure the concurrency level. Thanks! Cheers, Sergio B. -- Sergio Bossa http://www.linkedin.com/in/sergiob _______________________________________________ tc-dev mailing list tc-dev@lists.terracotta.org http://lists.terracotta.org/mailman/listinfo/tc-dev _______________________________________________ tc-dev mailing list tc-dev@lists.terracotta.org http://lists.terracotta.org/mailman/listinfo/tc-dev
