Re: MySQL ndbcluster - Man, why is it so slow!

[Robert Dye]

Thanks for the reply. I have attached my current config.ini, updated since reading your thread:

[NDBD DEFAULT]
ODirect
NoOfReplicas=2
DataDir=/mysql/data
BackupDataDir=/mysql/backup

SharedGlobalMemory=256M
DataMemory=1024M
IndexMemory=128M

FragmentLogFileSize=32M
NoOfFragmentLogFiles=64

LockExecuteThreadToCPU=0
LockMaintThreadsToCPU=1


MaxNoOfConcurrentOperations=100000
TimeBetweenGlobalCheckpoints=1000
TimeBetweenEpochs=200
DiskCheckpointSpeed=10M
DiskCheckpointSpeedInRestart=100M

[MYSQLD DEFAULT]

[NDB_MGMD DEFAULT]

[TCP DEFAULT]
SendBufferMemory=2M
ReceiveBufferMemory=2M

[NDB_MGMD]
HostName=10.160.90.84 # IP address of this server

# Storage Nodes
[NDBD]
HostName=10.160.90.85 # IP address of storage-node-1

[NDBD]
HostName=10.160.90.86 # IP address of storage-node-2

# Setup node IDs for mySQL API-servers (clients of the cluster)
[MYSQLD]
[MYSQLD]
[MYSQLD]
[MYSQLD]


What I have noticed is a slight performance increase for inserts, but heavily increased performance when doing updates. Although, MyISAM still is doing approx 3500 Inserts NDB is ranging only slightly higher with 1750 queries per second. I wonder what the magic button is at this point.