Decreasing Checkpoints & allowing for heavy load

[Adam D]

Ive been looking at tuning a few values in the cluster; Ive found two which look to be 'the ones' which will allow the cluster to carry out many querys, many updates etc. However not sure as to how to configure them.

2005-09-26 15:45:56 [MgmSrvr] INFO -- Node 11: Local checkpoint 822 started. Keep GCI = 821860 oldest restorable GCI = 821967
2005-09-26 15:57:37 [MgmSrvr] INFO -- Node 11: Local checkpoint 823 started. Keep GCI = 822181 oldest restorable GCI = 822293


NoOfFragmentLogFiles = 16
Of which defaults to 8, so I thought I would double this value. To provide 1gb of logfiles.

TimeBetweenLocalCheckpoints = 22
I was a little unsure of this one, as it mentions busy clusters will need to checkpoint right after another completes, and setting this to 6 or below would make this happen, however from what I read it seems that setting this higher would allow more write operations.


I was also unable to change some of these values with a rolling restart of the nodes, So Im not sure if this one will require a full cluster backup restore or a rolling initial for each node.

Can someone provide some more clarification here as to how to tune these values. Or if there are other values that I should look at to avoid issue when the cluster gets 'hit hard' with load.


Chopped up config;
MaxNoOfConcurrentOperations=170000
MaxNoOfConcurrentTransactions=150000
MaxNoOfLocalOperations=170000
MaxNoOfAttributes = 500000
MaxNoOfOrderedIndexes=500
MaxNoOfUniqueHashIndexes=500
##MaxNoOfFiredTriggers Default is 4000
MaxNoOfFiredTriggers=2000
MaxNoOfTables=350

TimeBetweenLocalCheckpoints = 22
TimeBetweenGlobalCheckpoints = 1500

[MYSQLD]
Id=2
hostname=172.16.250.1
BatchSize=256
MaxScanBatchSize=2M


Adam