[SOLR-6816] Review SolrCloud Indexing Performance.

• Type: Task
• Status: Open
• Priority: Critical
• Resolution: Unresolved
• Component/s: SolrCloud
• Assignee: Unassigned
• Reporter: Mark Miller
• Created: 03/Dec/14 23:06
• Updated: 12/Apr/15 18:23

Description

We have never really focused on indexing performance, just correctness and low hanging fruit. We need to vet the performance and try to address any holes.

Note: A common report is that adding any replication is very slow.

Issue Links

• relates to: SOLR-7344 Allow Jetty thread pool limits while still avoiding distributed deadlock.

Sub-Tasks

• The sync on the VersionInfo bucket in DistributedUpdateProcesser#addDocument appears to be a large bottleneck when using replication.: Closed
• It's difficult to saturate a Solr servers resources when bulk indexing. Using multiple JVM's on a single machine can drastically improve indexing performance.: Open
• Bulk loading with the default of updateDocument blocks all indexing for long periods of time.: Open
• Seed version buckets with max version from index: Closed
• Make the poll queue time configurable and use knowledge that a batch is being processed to poll efficiently: Closed