Indexing on different machine than your production machine

• Article Type: General
• Product: Aleph
• Product Version: 17.01

Description:
We would like to do our indexing on a different machine than our production machine. We intend to first test p_manage_01, p_manage_05, and p_manage_32. We will do the indexing this week, we would not do the actual test until after April 17th. For the test we would clone the database a second time on our test machine and try to import the indexing tables into it to see if it works, the details follow.

Our overall plan is:

1. Do a cold backup of the production database. At this time we would set up a trigger machanism to make a backup copy of all Z07 records created in SYS01. This would be the same type of triggers that are used in the 'Indexing the delta' that was done in upgrading from version 15.5 to version 17.

2. Clone the database to a different machine.

3. Run the particular indexing jobs we want to update on the clone of the database.

4. Export the indexing tables from the cloned database and copy them to the production machine.

5. Shutdown the production system.

6. Import the new indexing tables into the production machine and update any needed Z52 entries.

7. Copy the backup Z07 entries created by the triggers to the Z07 table and deactivate the triggers.

8. Startup the production machine with the new indexing tables in place, letting ue_01 reindex all of the new and updated records that were made while the indexing took place on the test machine.

My main questions are:

1. Does this sound like a plausible plan? Are there any holes in it or things that I missed?

2. Are there any indexing tables that do not get updated by ue_01?

3. What tables and Z52 entries are needed for each indexing job? The following is my list, is it correct?

p_manage_01
tables - Z95, Z97, Z98
Z52 entries - last-word-number

p_manage_102, p_manage_02 combo
tables - Z01, Z02
Z52 entried - last-acc-number, last-long-acc-number

p_manage_05
table - Z11

p_manage_07
table - Z13

p_manage_27
table - Z101

p_manage_32
table - Z0102

p_union_01, p_union_02
table - Z120
then create Z127 and triggers

Resolution:
1. Yes. It sounds like a plausible plan. You may want to look at the Parallel Indexing document. It covers some of the same topics.

2. All the indexing tables get updated by ue_01. KB 8192-2329 discusses this.

3. All of the indexing jobs read the last-doc-number (but, of course, do not update it).

p_manage_01: You need to consider the z980. When you copy the z95, z97, z98 back to the production machine, the prod z980 needs to dropped/created with util a/17/1.

p_manage_07 can, optionally, do the z00r. I presume that you are not using this (or do not need it to be recreated).