What is the impact of changing the Index level?

[Shane_Seagreen_]

Ok, following on from my thread regarding phrase searches, management are now considering going with full indexing instead of medium.

Besides the increase in database size, does anyone have any idea what impact on performance this will have on the server? i.e. Will the server need to completely rebuild the index (and if it does, does this make the search facility unavailable?), or is simply a partial rebuild?

[TonySterling · Accepted solution]

Going from Medium to Full will be around a 4% increase in size for the index. If you make the change, going forward you can phrase search, but the index will have to be rebuilt to be able phrase search the entire archive.

here is the blurb from the documentation.

Indexing level: Use this to specify how detailed an index Enterprise Vault is to create by default. The more detailed the index, the more complex the searches that are possible using that index. However, more detail also makes the indexes larger, so you have to balance the index size against the amount and type of searching that you expect users to do.

Brief The index created by Enterprise Vault will enable searches on the following attributes of each item: Author, Subject, Recipient, Created Date, Expiry Date, File Extension, Retention Category, Original Location. If a search matches an attachment to an item, the search result contains both the main item and the attachment.

Medium As for Brief, and in addition enables searches on the content of each item, excluding phrase searches.

Full As for Medium, and in addition enables searches for phrases in the content.

The size of the index increases going from Brief to Medium, and from Medium to Full. It is not possible to give an exact size for the index because the size depends on the data that is indexed. However, the different settings will probably behave as follows:

Brief will give an index that is approximately 3% of the size of the data that is indexed
Medium will give an index that is approximately 8% of the size of the data that is indexed
Full will give an index that is approximately 12% of the size of the data that is indexed

[TonySterling]

And yes, while the index rebuilds it will not give you complete results. How large is it? You could start the rebuild on a Friday and let it run over the weekend.

[Lee_Allison]

The new indexing level takes over from the point you make the change going forward. So for all items in an archive to have the new level, yes, you need to rebuild the index.

You found the estimates on index size for full indexing? The indexes should take up approximately 12% of the total archive size, just fyi.

[Scott_Anderso1]

We're using full indexing and it's approx. 9% of the total data.

[Lee_Allison]

Scott, then your people might be archiving a lot of jpg's or other file types with little or no indexable data.

From the docs:

Estimated size of indexing data
Indexing level Estimated size

Brief 3%

Medium 8%

Full 12%

[Micah_Wyenn]

Yeah, kids, remember it's a range (usually 8-12%), not 12% automagically. When capacity planning, shoot for the worst case (if you're doing PERT) at 12% and best at 8% and you should have a good range for your top value.

micah