Bug #25472 Patch: Speed up execution of NDB Cluster when using Dolphin Super Sockets
Submitted: 8 Jan 2007 18:24 Modified: 15 Dec 2010 14:04
Reporter: Lenz Grimmer Email Updates:
Status: Verified Impact on me:
Category:MySQL Cluster: Cluster (NDB) storage engine Severity:S4 (Feature request)
Version:mysql-5.1 OS:Any
Assigned to: CPU Architecture:Any
Tags: Contribution

[8 Jan 2007 18:24] Lenz Grimmer
The attached patch was contributed by Mikael Ronström on behalf of Dolphin Interconnect.

This patch introduces 2 new configuration parameters in MySQL Cluster that is very effective in speeding up execution when using Dolphin SuperSockets and should in many cases also be of great value in other scenarios as well.

 - SchedulerExecutionTimer
 - SchedulerSpinTimer

SchedulerExecutionTimer means that the scheduler will check for more data from
the sockets before it will send the socket. This means in general larger packets sent and thus higher throughput. Potentially it can increase response times. The
parameter is set in microseconds.

SchedulerSpinTimer means the minimum time that the scheduler should wait for more data on sockets without going to sleep. Setting this to 10000 or higher would mean that the main thread of the NDB data node never sleeps. It is provided in microseconds.

Both parameters can be set also while MySQL Cluster is running through the management client command.

How to repeat:
Apply the patch, observe the new functionality.
[8 Jan 2007 18:25] Lenz Grimmer

Attachment: dolphin_sched.patch (text/x-patch), 19.96 KiB.