Improving performance Percona Mongo v3.6.17-4.0 standalone

MongoDB Percona Server for MongoDB
1

Hi there,

On the daily basis, from a ReplicaSet a dump is made, we start a standalone server, removing any replicaSet configuration, on that standalone server a Java application will sanitized/scrambling personal data like lastnames, e-mails, etc. The Java application uses multithreading. Running some updateMany over some collection takes many hours, we can have around 300K e-mails to be sanitized/scrambling. We have a “kind of big” collection of 45GB “73 MM of documents on that collection and many indexes”, the entire database is about 91GB, the Debian 8 box has 14GB of RAM.

From Mongo config, I have setup a bit more of cacheSizeGB to try to improve some performance. I was wondering what other settings I can modify to try to get performance?

wiredTiger:

engineConfig:

 cacheSizeGB: 10

checkpointSizeMB: 1000

statisticsLogDelaySecs: 0

journalCompressor: snappy

directoryForIndexes: false

collectionConfig:

blockCompressor: snappy

indexConfig:

prefixCompression: true

The next setting to try will be to disable the journal, and see if it will increase some performance.

Server: Percona Mongo v3.6.17-4.0 standalone

Thanks in advance.

2

Hi, disabling the journal will help in terms of performance but be careful about the durability. Are you using the unordered bulk write functionality? that should be faster as well:

https://docs.mongodb.com/manual/reference/method/db.collection.bulkWrite/#bulkwrite-write-operations-updateonemany