CA Service Management

Hello Team,

We have almost 20K's CI's which were being processed each day using GRloader and it almost takes 2-3 hours to complete the entire process.

Is it common that GRLoader takes this much time to update the records and to be more specific it takes much time to load the data from TWA to CMDB than from csv/xml to TWA.

Otherside, what are the best practices to have an integration/automation to load the CI's from XML into CMDB daily.

Regards,

Sharath.M

Hi Saranth,

Yes, GRLoader will take time in loading that many Cis due to the checks/reconciliation that takes place during the process.  Unfortunately there really isn’t a “speedy” way to import Cis and ensure that duplicates are not created at the same time.

Jon Israel

Principal Support Engineer

Thanks much Jon for a speedy answer.

Yes, I thought of, but we have got a new requirement which almost have 30K CI's and we should process them each day.

Wondering, how the integrations with external applications/csv files is being done with other clients with CA CMDB.

In 30K, only 10% will be changed on a daily basic but its NOT easy for the client to send CA only the cumulative data

Regards,

Sharath.M

Hey Saranth,

Using -slump switch may help to a certain extent. It uses direct communication than the webservice calls

-jobby

Hi Sharath,

To add what Jobby had mentioned, when grloader is used, CORA is used to register the CI's. CORA has it's own rules about registering assets and it makes an asset unique. So the delay may be caused by these API calls to check uniqueness of the CI. The more assets/rows there are in the ca_owned_resource table, the slower grloader will be as CORA needs to reconcile two columns in every row of that table to make sure no duplicate assets are created. Best approach is to use -slump switch (only if URL is from primary server). You may also type grloader -h command to know more details about the switches.

Thanks,

Naveen

Thanks Jobby and Naveen.

I will try with the Slump switch and update the results.

Thanks again for your suggestion.

Little outside to original concern..

When I use GRloaded it works fine but when I go with "Slump" switch it puts the below error in the logs..

ERROR grSlump 1320 GRSlump5100 Unable to fetch object:  factory(mfrmod) where(id = U'C49E8C33B593874A874C63D486DF5FBB') reason:Attribute persistent_id not found by producer ca_cmpny in dotted reference manufacturer.persistent_id

grCI 1705 GRCIDIF-0700 SREL Lookup of attribute(model) - relattr(id) value(C49E8C33B593874A874C63D486DF5FBB) lookedupValue(null)

I have this model and Company(manufacturer) created properly, any thoughts on this..

Further, in TWA if I remove the model and Manufacturer  and  get it executed even with "Slump" switch.

Thanks,

Sharath.M

HI Sharath.

How does your input look like. Can you give us an example, how do you specify model and manufacturer in your input for grloader?

Thanks and regards

..........Michael

HI Michael,

We have imported the CI’s from an xml to TWA and made the records status to Ready and then executed the below command. Manufacturer and Model were already imported to TWA from xml along with other attributes

grloader -lftwa -u ServiceDesk -p XXXXX -a -n -P -I –slump

Thanks

Sharath.M

A completely different way to look at this is to only load the incremental data, meaning only load the differences.

You would need to build some logic for that. Probably a way to export the same XML structure as your input from CMDB, then a way to compare the XML and create a resulting XML that only contains the differences.

Not sure how much faster that will be, but when written in a performant way I think this can save you some processing time.

Hi,

True, we were trying to use BIZtalk to process this but even its not feasible to make only the difference data from BizTalk.

Was looking some other options to make only the new data

Sharath.M

CA Tools Lead

+46 73 4184282

sharath.mateti@saabgroup.com<mailto:sharath.mateti@saabgroup.com>

<http://www.techmahindra.com/>

Connect with us:

<http://www.facebook.com/TechMahindra>[cid:image005.jpg@01CEC690.43D62120]<http://twitter.com/Tech_Mahindra>[cid:image006.jpg@01CEC690.43D62120]<http://www.linkedin.com/company/tech-mahindra>[cid:image007.jpg@01CEC690.43D62120]<http://www.youtube.com/user/techmahindra09>

How about trying to split the data into several chucks? Instead of trying to load all 30K at once, perhaps split it into 5K - 10K records per load and the execute these at different times over night?

Regards,

Brian