nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

DataDeploy Replicant performance issues

fish

I've been having some performance issues with DataDeploy with certain DCRs that have replicants since upgrading some time ago. It is getting on my nerves and now and I wish to do something about it ...

DataDeploy works fine until I have a DCR with a nested replicant. Then I get a performance issue (a Java process grabs a CPU for 3-4 minutes and finally DataDeploys the DCR). I get no errors and everything appears OK. I only get the problem with DCRs that have nested replicants.

This is happening for DAS and manual deployments.

Has anyone else come across this?
Am I missing something obvious?

TS 552. DD 5.6 latest patch level. Solaris/Oracle

Grateful for any input,

Steve.

Find more posts tagged with

Comments

Migrateduser

we had a problem with the real-updates setting in our database.xml file...here is what tech support said:

If real-update is set to "yes", DD assumes that user will not modify PK values and proceeds to perform "real updates", (i.e.) executes a series of UPDATE SQL statements. Otherwise, existing rows are deleted first followed by inserts.

Since setting real-update to "no" solved your issue, then I suggest you go with that. There are few know issues with setting real-update to "yes" especially when handling replicants.

-------------------
World of Warcraft  Illidan/Horde

fish

Thanks for that but our real-update are set to "no" in all our dd.cfg files.

Have you got instances of nested replicant DCRs that are 'datadeploying' without any performance issues?

Steve.

Migrateduser

all of our website ads are handled via 1 huge dcr w/ nested replicants...it has about 15-20 and takes maybe 30-35 seconds to deploy (im guessing due to all the database transactions)

-------------------
World of Warcraft  Illidan/Horde

Migrateduser

Hi Steve,
Is the performance issue with or without the "row-map-cache-file" option in the dd config.

Thanks
srini

fish

Hi Srini,

You could be onto something here...
The 'row-map-cache-file' option is not set.
It sounds like a good starting point in resolving the performance issue.

13-00 update
Issue Resolved many thanks Srini

-----------------------
Steve.

Edited by fish on 01/11/05 05:32 AM (server time).

ChiragK

The Real issue with replicant is that, when data deploying, DD creates a cartesian product of the nested replicant. Now the problem is if you have too many nested replicant the cartesian product is a huge number (imagine you want to update only couple of hundred records, but due to cartesian product data deploy expect probably hundreds of thousands to a million row update). Therefore you have high CPU consumptions and also a large pool of jvm memory wasted.

Check the need for the replicant size in the DD CFG. Try to reduce unwanted replicant size and you will see a large difference in performance.

Hope this helps !!!!!

fish

Many thanks - that is useful information.

I have now fixed the issue with the "row-map-cache-file" attribute.

Thanks again,
Steve.

-----------------------
Steve.

nickcolorado

Hi,

It sounds like I am having a similar issue with datadeploy. Do you know where the documentation for row-map-cache-file setting is? I did not see it in the OD Database Admin guide.

Thanks for your help!
Nick Apel