we are planning to Delete the Failed Instances by deleting the respective reports which caused it

[former_member599993]

We have identified around 8.5 Million Failed instances on our platform and also found that these instances belong to a small set of reports, SAP engineer asked us to delete the reports which will automatically clean up the instances from FRS and also will correct the metadata on CMS DB but we would like to know if this creates any performance impact on the Platform. We are interested to know how different is this process from running the Instance deletion script. We can estimate the performance impact if we know the backend process which runs to clean up these instances when we delete the respective reports

[Joe_Peters]

Why not just set a low instance limit and let the purge process take care of them?

[former_member599993]

Hi Peters,

Yes, eventually will be following the approach of Limiting Instance count at folder level but it's a time taking process as it requires multiple levels of business communication so we are trying to mitigate this issue meanwhile to improve the overall system performance.

[Joe_Peters]

Changing the instance instance limit on a few reports should be less impactful than deleting them, and should have less impact on performance to boot.

[former_member599993]

Is there any difference between the back end process which gets kicked off when you set the instance limit Vs when you delete the report?.

Just curious to know how the Performance impact will be minimal when we set the Instance limit. Assuming it will follow the same process to delete the instance history from FRS and CMS DB in the same way

Thanks!

[Joe_Peters]

When you delete a report in CMC or launchpad, the report and all child instances are immediately removed from the CMS database and FRS. When you change the instance limit, the child instances are removed either upon the next scheduled run of the instance, or (I believe) once per day. I've heard that this purge process is very low priority and as such would have less impact on performance.

[denis_konovalov]

CMS and FRS services will be involved in this process.
CMS will be running a lot of relationship queries and FRS will be the service actually deleting instances from disk location.

So, your CMS DB will be tasked with lots of queries, CMS service will eat a lot of memory and disk I/O will be taxed.

Impact depends on the sizing of your system and how well it can handle those types of loads.

[former_member599993]

Thanks Denis, Appreciate your swift turnaround as always!

so it's clear that the CMS services and the system resources on those nodes will be heavily loaded during this activity considering the Huge number of instances we have.

I am inclined towards testing this approach on our lower environments to check the capacity of platform to handle this load, in that way we can estimate the resource consumption and the number of instances it deletes for certain interval of time