Solr sharding DB_ID_RANGE (node deletion & shard fragmentation)

[luis_alventosa]

Hello,

I am looking at ASS 2.0 Solr sharding methods and wondering how DB_ID_RANGE handles node deletions.

In my current usage of Alfresco I inject thousands of new documents per day but I also delete thousands of documents per day.

I like the advantages of DB_ID_RANGE vs DB_ID because it will help me scale easily as my index grows without having to reindex but as DBIDs are not recycled by Alfresco this would mean that my shards may end up using only 50% or maybe less of their capacity.

Could someone shed some light on this?

Which are the scenarios where DB_ID_RANGE is suitable?

Lots of node creations and very few node deletions?

Thanks a lot,

Luis ALVENTOSA

[angelborroy · Accepted solution]

You're right. DB_ID_RANGE is not handling document deletions.

You may use DB_ID_RANGE with a dynamic resource setting. I mean, you may allocate resources (CPU & RAM) according to the real number of documents in each Shard.