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

Ask Your Question
0

Questions on paritioning distribution in cluster

I have Gigaspaces setup in a 2 machine cluster separated into two data centers that communicate with a WAN connection. So that's one machine per data center.

I have one partition and want 2 primary PU instances with a backup for each instance. So that's a total of 4 instances (2 primary and backups for each).

I have setup 2 GSCs per machine.

I prefer all my primary PU instances on one machine and the backups to be on the other machine. I don't really care which machine has the primary instances so long as they are all on that machine and the backup instances are on the other machine.

The reason for this is that the WAN is obviously slower than local communication so it makes sense to have all your primaries in the local LAN (and in my case on the same machine).

I looked at http://wiki.gigaspaces.com/wiki/displ... and didn't really find a suitable answer.

TIA.

asked 2013-10-31 15:09:42 -0500

Steve McCarthy gravatar image

updated 2013-10-31 15:17:27 -0500

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted
0

Steve, I believe you are looking for this: http://wiki.gigaspaces.com/wiki/displ...

Shay

answered 2013-11-01 14:44:46 -0500

shay hassidim gravatar image
edit flag offensive delete link more

Comments

I see, so I start the GSCs in one data center with -Dcom.gs.zones=zone1 and the GSCs in the other data center with --Dcom.gs.zones=zone2.

Then I configure this in the SLA portion:

<os-sla:sla number-of-instances="2" number-of=backups="1" max-instances-per-zone="zone1/1,zone2/1">
    <os-sla:requirements>
        <os-sla:zone name="zone1" />
        <os-sla:zone name="zone2" />        
    </os-sla:requirements>
</os-sla:sla>

Does that sound correct?

Steve McCarthy gravatar imageSteve McCarthy ( 2013-11-01 16:52:49 -0500 )edit

That's the idea Steve.

It is not recommended having primary and backups in different data centers that are more than few miles apart due-to the latency impact and the failover behavior. Primary and backups must be in the same data centers. This provide maximum performance and reliability.

If you want distributed deployed across multiple data centers to share data between the,, each data center should have a separate grid started and its own data grid with primary-backups deployed with a WAN gateway replicating relevant data between the data centers.

With this approach there would not be any performance impact when the primary replicating its data to the backups. See: http://wiki.gigaspaces.com/wiki/displ... - Multi Master http://wiki.gigaspaces.com/wiki/displ... - Multi Slave

Shay

shay hassidim gravatar imageshay hassidim ( 2013-11-01 17:13:57 -0500 )edit

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

Stats

Asked: 2013-10-31 15:09:42 -0500

Seen: 83 times

Last updated: Nov 01 '13