[DOCS-402] Sizing considerations for cluster Created: 15/Aug/12  Updated: 15/Nov/21  Resolved: 15/Aug/12

Status: Closed
Project: Documentation
Component/s: None
Affects Version/s: None
Fix Version/s: None

Type: Task Priority: Major - P3
Reporter: Yash Badiani Assignee: Unassigned
Resolution: Done Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Participants:
Days since reply: 11 years, 27 weeks, 1 day ago

 Description   

I am currently working on sizing a mongodb cluster. I am taking into the following considerations for sizing. Please let me know if i missed any:

1) Anticipated Size of the data(Number of documents * Size of documents * Number of users)
2) Expected size of data to be replicated
3) Size of Indexes(for each collection,5% of data size if default "_id" index, 5% for each additional index on each collection)
4) Size of the working set(to determine size of RAM)
5) Data archival requirements

Let me know if there are additional things i need to consider.

Also, is there a guideline on the number of shards to be designed for a cluster?

Thanks,
Yash



 Comments   
Comment by Sam Kleinman (Inactive) [ 15/Aug/12 ]

Thanks, I think the google group will provide a more thorough answer to your question. Additional things to consider:

  • MongoDB also uses some "padding" that increases storage requirements to support documents that grow after creation.
  • The size of indexes is more variable, but that depends on the size of the keys.
  • In general you can have anywhere from 2 shards to a few hundred. I'm not aware of any deployments larger than that at the moment, but there's no reason that there couldn't be more than that. Basically as many as you need.
  • The main sizing recommendation for shards is: ensure that you're, at any moment, using something less than 80% or so of total capacity (write capacity, storage capacity, etc.) Basically it takes some amount of time and system resource for the cluster to redistribute data (i.e. to "balance") when you add a new shard and if you wait too long to add capacity, it may be difficult or effectively impossible to reach balance without some kind of downtime.

Hope this helps!

Comment by Yash Badiani [ 15/Aug/12 ]

Will add this to the google group as this is more of a query and not a bug. Sorry for that.

Generated at Thu Feb 08 07:38:36 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.