Working set Mongo 3.0.11

79 views
Skip to first unread message

MsTech

unread,
May 4, 2016, 3:03:04 AM5/4/16
to mongodb-user
Trying to add index size plus data size to get working set, but I have the raw data values of each but I need to combine them in a script. Has anyone seen a script like this that can calculate working set? I pull all the db.stats() but I still need a script to calculate working set, or at least some suggestions on how to start. It seems like there some be a standard script out there because everyone would like to have this information. Thx in adv

Kevin Adistambha

unread,
May 9, 2016, 11:43:12 PM5/9/16
to mongodb-user

Hi,

Trying to add index size plus data size to get working set, but I have the raw data values of each but I need to combine them in a script. Has anyone seen a script like this that can calculate working set? I pull all the db.stats() but I still need a script to calculate working set, or at least some suggestions on how to start. It seems like there some be a standard script out there because everyone would like to have this information. Thx in adv

In MongoDB terms, "working set" refers to the data that is most often used in the database. Generally, this includes all of the indexes, and frequently-accessed documents. Please see this page and also this page for more information regarding working set.

Therefore, calculation of a server’s working set is very use-case-specific, and highly dependent on the nature of the data and its operations. Such as:

  • whether the queries performed on the database are targeted at specific subset of the data, or large portions of the data
  • the nature of the workload (read-heavy, write-heavy, or somewhere in between)
  • document sizes
  • index sizes, and the number of indexes
  • the storage engine used

In general, your server should have enough RAM to hold the entire working set in memory, so that MongoDB spent as little time as possible moving data to and from disk. For more information, please see Optimization Strategies for MongoDB.

Best regards,
Kevin

Reply all
Reply to author
Forward
0 new messages