AnsweredAssumed Answered

GRLoader & Performance

Question asked by dconnell.1 on Sep 17, 2015
Latest reply on Dec 2, 2015 by EricoDellaVallePrezzi76043714

I am new to SDM and GRLoader.  So sorry for my ignorance.

I have 4 questions below.

 

We have potentially up to 500,000 Software CIs and 500,000 relationships to load using GRLoader from BMC ADDM (Discovery Product).

This is a new feed.  I am designing the interface.

 

Existing tools (such as ITAM) are currently loading CIs at about 1,000 in 12 minutes. 

This means it will take 9 days to load everything from ADDM.

 

After the initial load, I could speed up subsequent updates, if I pre-retrieve the Asset ID, (which I think  is own_resource_uuid)

from the ca_owned_resource table by looking up the unique CI name and use the Asset ID as the id in GRLoader.

I would plan to load CIs whether they have changed or not (to simplify the code - so I don't have to calculate Deltas).

 

  • A) If I specify the ID in GRLoader for all pre-existing CIs, how much faster will GRLoader run, since I will effectively be bypassing CORA?

 

From other staff here at the Bank, I have understood that CORA creates duplicates at a high frequency - mostly because Serial Number

is not reliable in a Virtualized world and is often set to null or ADDM flip flops due to discovery "issues".

so bypassing CORA has some attractions.

 

  • B) Any thoughts?  Any other design options?

 

Also, would I gain any performance increase if I loaded TWA using SQL rather than via GRLoader.  I have tested this and it works.

I got hints from a CA Powerpoint deck that loading the ci_twa_ci table directly was OK and did not break referential integrity.

 

  • C) Is it safe to load the TWA table directly?

 

  • D) I suspect the slow part is loading TWA data into the CMDB.  Is this true?

Outcomes