Solutions

MarkLogic Data Hub Service

Fast data integration + improved data governance and security, with no infrastructure to buy or manage.

Learn More

Learn

Stay On Top Of Everything MarkLogic

Be the first to know! News, product information, and events delivered straight to your inbox.

Sign Me Up

Community

Stay On Top Of Everything MarkLogic

Be the first to know! News, product information, and events delivered straight to your inbox.

Sign Me Up

Company

Stay On Top Of Everything MarkLogic

Be the first to know! News, product information, and events delivered straight to your inbox.

Sign Me Up

 
Knowledgebase:
How in-memory cache works in MarkLogic Server
26 April 2016 02:52 PM

Summary

The following article explains the way in-memory caches are used by MarkLogic Server and how can they be utilized to improve query execution.

 

Detail

MarkLogic Server provides several caches that are used to improve the performance during query execution. When a query executes for the first time, the Server will populate these caches to store termlist and data fragments in memory.

MarkLogic Server keeps a lot of its configuration information in databases, and has a lot of caches to make it run faster, but those caches get populated the first time things are accessed. The server also uses book-keeping terms in the indexes to keep track of whether all documents have been indexed with the current settings. MarkLogic caches this information, but has to query the indexes on the first request to warm the cache.

The in-memory cache in MarkLogic Server holds data that was recently added to the system and is still in an in-memory stand; that is, it holds data that has not yet been written to disk.

For updates, if there is no in-memory stand on a forest when a new document is inserted, the server will create it. This stand is big enough for thousands of documents, but the cost of creating it will be seen in the time taken for the first document added to it.

 


How will the in-memory cache help improve query execution

When a query is executed, the in-memory data structures like range indexes and lexicons get pinned into RAM the first time they are used.  The easiest way to speed things up is to "warm the caches” by running a small sample program that exercises the type-ahead prior to starting production. You can also keep the server warm by doing a non-time-critical stub update at time intervals (every 30 sec to 1 minute). If the server is idle, then it will serve to keep caches and in-memory stand warm. If the server is really busy then it would only take a small amount of extra work. Once this is done, the functionality will be fast for all users in all future sessions.

(2 vote(s))
Helpful
Not helpful

Comments (0)