Currently my domain has 2 databases which are nearing their 65Gb limit on Windows. If am correct, I have 3 options:
Clear out as much of the database as possible and keep it below 65Gb (not possible)
Create 2, or more, databases and split the information between the new databases
Move to a UNIX based OS and use the larger file size limit.
So my question is, out of options 2 or 3, which is better for server performance? Access to the many smaller db’s wouldn’t be an issue, as there is an automated entry system which would take the user to the relevant database, and there is no interraction between the documents in the database.
By using say 5 smaller databases instead of 1 large one, can anyone comment on whether the server load would be greater or less, or the same?
In terms of performance, it’s mostly how the application is used. If people use their own department/region and most of your users at any one time are from one dept/region, then splitting has great performance benefits.
If people frequently need to go across boundaries, then you’ll probably take a small hit.
The answer to your question would be ‘It depends’. Actually a combination of 2 and 3 is probably the answer. In evaluateing the merits of splitting the DB will depend to a large extent on how people use the DB. For example when they go to a document are they likely to open the document and then the attachment? Or do the look at the document and periodically the attachment? Are there frequent changes to the document and/or attachments?
If they primarily look at the document and perhaps make changes to it but don’t look at the attachment very often then create an action that acts on new documents that moves the attachments to a second DB and link the ‘Parent’ to the ‘Child’ that contains the attachments. This will greatly speed up opening the original document because the attachments are only uploaded when required.
If both the Document and the attachments are opened and changed regularly then having them all in the same DB might make sense. However, perception is everything from the end users point of view. If you significantly improve the speed to open the ‘Parent’ document and stall the time for opening the attachment until they click on the link, they will likely percieve that the system is working faster.
In my opinion, the server load will be greater, because it will have to manage more indexes, and more db-connections (related to the log, the filesystem access, etc etc…).
But, considering that the server load depends mainly on the documents’ sizes, and the index-type (attachments or not, etc…), the difference could be not so huge as that.
Thinking of that, moving to a Unix based system, and splitting the DBs seems to be the better solution (FS more efficient, DBs file size limit, …).
Thanks for the replies, I’ll check out the red book.
The database holds documents which are copies of emails, so they will range from a few 100kb to appox 20Mb (don’t ask). Many of them will have more than 1 file attachment.
Users will jump between databases quite frequently and I imagine all the databases would be held on the same Domino server, so going on what you’ve all said, I might get a drop in performance using more smaller databases
