Welcome to the new Gigaspaces XAP forum. To recover your account, please follow these instructions.

Ask Your Question

High CPU usage

We have jboss server running three embedded spaces on a Solaris 10 Sun-Fire-V445.When a space operation is performed, CPU usage is high -35-50%.. The operations are simple reads/writes to space of a small amount of data, sequential and low volume. I tried the benchmark available through the GS Management Center UI using the defaults on a PC, and while the benchmark completed quickly, the CPU usage went up to 83% (See attached). Do you suggest any tuning.

What are the requirements for space to run in terms of memory: 1. Memory footprint of a single GS instance when it starts up; and assuming no data in it. 2. Memory needed to cache a single java object instance of size x. I will configure jboss accordingly.

Is there a Gigaspaces XAP System requirements doc that specifies what is required (and recommended) in terms of cpu, memory, network etc.


Edited by: s m on Apr 17, 2008 6:07 PM h4. Attachments


{quote}This thread was imported from the previous forum. For your reference, the original is [available here|http://forum.openspaces.org/thread.jspa?threadID=1915]{quote}

asked 2008-04-17 16:57:17 -0500

wasefmasood gravatar image

updated 2013-08-08 09:52:00 -0500

jaissefsfex gravatar image
edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted

If the space is not doing anything , it should not consume any CPU. There are few heartbeat mechanisms going on between the client/space , space/space and client,space/LUS that will have very minimal impact ("white noise") - but these are negligible in terms of CPU consumption and network utilization.

In general, Embedded space activities takes more CPU. The reason is no IO done, so the machine CPU is fully busy all the time with executing business logic and space operations without "forking" IO operations.

The Space Memory footprint (empty one) is very minimal. With 32 bit JVM it would be less than 10M. With 64 bit JVM it will be 20M. This also includes a LUS running within the same JVM.

The footprint of a space object depends with:
- Space Class amount of indexes and their values variance - The more you have the footprint will be increased.
- Space Class fields amount - The more you have , the space object that actually stores your object data will have more meta data associated with it.
- Space Class values footprint - The bare bone object field's data footprint. By default these are not compressed when stored within the space.

The LUS footprint is depends with the amount of clients performed lookup.

To minimize the footprint of the space objects you should use the binary serialization pattern.
See wiki page:

The following will help you to implement the above:

For additional space performance optimization suggestions see:


answered 2008-04-18 06:22:48 -0500

shay hassidim gravatar image
edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Question Tools

1 follower


Asked: 2008-04-17 16:57:17 -0500

Seen: 1,134 times

Last updated: Apr 18 '08