Setting Up Deferred Indexing

When you set up deferred indexing, you choose which indexes to process immediately and which to defer. Deferred indexing provides two main features that aid in the processing of your indexes:

A system of queues. A queue is a list of records and indexes waiting for processing. You assign each index to a queue by assigning the index a number (for example, 0 or 1. By default, all indexes are assigned to queue 0). Horizon processes indexes assigned to queue 0 as soon as your library staff saves or imports a MARC record. Indexes assigned to queue 1 are deferred. This means that Horizon places each saved or updated index in a queue to be processed on a separate workstation.

Important: Although the system allows up to 255 queues, SirsiDynix strongly recommends you start with a single deferred queue. (For information about setting up multiple queues, see Setting Up Complex Deferred Indexing .)

Queue Indexer. The queue indexer is the utility that processes the indexes you choose to defer. Horizon’s queue indexer is called “QIndex.” This is a command-line utility program that runs on a computer you dedicate to deferred indexing.

QIndex works by reading a series of rows from the mq_index_queue table and then processing these entries on a record-by-record basis for a queue. The mq_index_queue table is a list of records that QIndex needs to index. After a record has been indexed, QIndex deletes the queue rows for that record (the deletion is part of the same database transaction as the indexing itself).

You defer processing of specific indexes by ranking indexes on level of necessity. In general, you can group your indexes into these two categories:

Non-deferred indexes. These are the indexes you process immediately. These indexes are used by your cataloging or acquisitions staff to find and manipulate the record after it is saved. Non-deferred indexes are processed on your technical workstation (where your cataloging transactions occur). Because these indexes are used by cataloging and acquisitions staff to complete the processing of the materials, they must be updated immediately.
Deferred indexes. These are the indexes where the processing is postponed. These indexes are either highly specialized or used for public searching purposes. Because these indexes are not immediately necessary, you can choose to defer them. Deferred indexes are processed on the QIndex workstation (this is the computer you have dedicated to deferred indexing). Because there is generally a delay between the time you import a record and when you make the item available to the public, a short delay in indexing should not be critical to the smooth operation of your library.

Here is an example of how deferred indexing works:

Workstation

Indexes

Time per Record

Technical Workstation

12 fundamental, non-deferred indexes

2 seconds

QIndex Workstation

68 specialized and public indexes

36 seconds

In this example, Horizon indexes the 12 fundamental, non-deferred indexes on your technical workstation when a record is saved. At the same time, the 68 deferred indexes are indexed on the QIndex workstation. Although QIndex will fall behind because it has more work to do, it can catch up during times when large amounts of MARC data is not being saved. Because QIndex operates on its own workstation, your staff is not kept waiting while QIndex processes the indexes.

This section explains these topics:

 


© 1998-2017 Sirsi Corporation