Slow cache updates with indexing module enabled

classic Classic list List threaded Threaded
1 message Options
xero xero
Reply | Threaded
Open this post in threaded view

Slow cache updates with indexing module enabled

We are experiencing slow updates to a cache with multiple indexed fields
(around 25 indexes during testing but we expect to have many more) for
updates that are only changing one field. Basically, we have a
customer*->belongsto->*segment relationship and we have one column per
segment. Only one column is updated with a 1 or 0 if the customer belongs to
the segment.

During testing, we tried dropping half of the unrelated indexes (indexes
over fields that are not being updated) and we duplicate the performance. We
went from 1k ops to 2k ops approximately.

We found these cases may be related:

Could you please confirm us if IGNITE-7015 could be related to this
scenario? If yes, do you have any plans to continue the development of the

We are using Ignite 2.7.6 with 10 nodes, 2 backups, indexing module enabled
and persistence.

Cache Configuration: [name=xdp-contactcomcast-1, grpName=null,
memPlcName=xdp, storeConcurrentLoadAllThreshold=5, rebalancePoolSize=2,
rebalanceTimeout=10000, evictPlc=null, evictPlcFactory=null,
onheapCache=false, sqlOnheapCache=false, sqlOnheapCacheMaxSize=0,
evictFilter=null, eagerTtl=true, dfltLockTimeout=0, nearCfg=null,
writeSync=PRIMARY_SYNC, storeFactory=null, storeKeepBinary=false,
loadPrevVal=false, aff=RendezvousAffinityFunction [parts=1024, mask=1023,
exclNeighbors=false, exclNeighborsWarn=false, backupFilter=null,
affinityBackupFilter=null], cacheMode=PARTITIONED, atomicityMode=ATOMIC,
backups=2, invalidate=false, tmLookupClsName=null, rebalanceMode=ASYNC,
rebalanceOrder=0, rebalanceBatchSize=524288, rebalanceBatchesPrefetchCnt=2,
maxConcurrentAsyncOps=500, sqlIdxMaxInlineSize=-1, writeBehindEnabled=false,
writeBehindFlushSize=10240, writeBehindFlushFreq=5000,
writeBehindFlushThreadCnt=1, writeBehindBatchSize=512,
writeBehindCoalescing=true, maxQryIterCnt=1024,
rebalanceDelay=0, rebalanceThrottle=0, interceptor=null,
longQryWarnTimeout=3000, qryDetailMetricsSz=0, readFromBackup=true,
nodeFilter=IgniteAllNodesPredicate [], sqlSchema=XDP_CONTACTCOMCAST_1,
sqlEscapeAll=false, cpOnRead=true, topValidator=null, partLossPlc=IGNORE,
qryParallelism=1, evtsDisabled=false, encryptionEnabled=false]


Sent from: