Hi Guys,

I have to ask a somewhat fuzzy question regarding write performance within 
a transaction.

I'm using the Document API (meanwhile) and a Document that has a few 
indices.

Within a transaction I'm update around 8000 document instances of a 
specific class.

When I commit the transactions it takes sometimes 15-25 secs. for the 
transaction to get commited, somtimes only a few seconds.

What could be the reason for such an inconsistent behavior?
Is the timing okay, from what your experience is?

I'm using index.txMode=FULL.
However removing that doesn't change things.

Also, I have the impression that accessing instances of the same class from 
another thread gets "stalled" as long the transaction is committing.
Is that "normal"?

The transaction above is one of the biggest and I guess in the face of 
other applications that use OrientDB it's not that big.
So, are there any settings that would be better for such smaller 
transactions than the default ones.


Currently I use:
-Dindex.txMode=FULL
-Dstorage.wal.cacheSize=10000
-Dstorage.diskCache.bufferSize=4000
-Dstorage.wal.restore.batchSize=25000
-Dstorage.wal.fuzzyCheckpointShutdownWait=30
-Dstorage.wal.maxSize=250

Thanks,
Markus

-- 

--- 
You received this message because you are subscribed to the Google Groups 
"OrientDB" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to