PDCR Job took more time to copy approximately same amount of data

Database
Highlighted
Teradata Employee

PDCR Job took more time to copy approximately same amount of data

I have observed for one of my customers that PDCR job took more time for approximately same amount of data to copy. For ex-

PDCR DBQL job took 17 Mins and 22 Secs to Copy 8.19 M records which is approximately close to yesterday’s 8.09 M records which took only 8mins and 28 secs.

 

i am not sure why it took almost double the time to copy same amount of data, where i have observed that there was no blocking or any performace issues, neither we have made any changes.

is it related to the kind of data that it is copying? is is there anything else which i am missing?

Tags (1)
3 REPLIES 3
Ambassador

Re: PDCR Job took more time to copy approximately same amount of data

Hi,

 

Your observed difference is in elapsed time and there are a number of things that can affect that (you've already discounted blocking which is one of them). How about:

  • throttling - were the queries in the 'long run' delayed?
  • priority - did both sets of queries run in the same WD?

If you've got DBQL for both sets of queries, look at AMPCpuTime, TotalIOCount and the 'physical IO' columns. These will tell you if the two sets of queries are doing the same amount of work.

 

HTH

Dave

 

Ward Analytics Ltd - information in motion
www: http://www.ward-analytics.com
Teradata Employee

Re: PDCR Job took more time to copy approximately same amount of data

Hi Dave

 

Thanks for the respose and suggestion, yes i have compared the result and can see more ampcputime,totaliocount and physicalio count for the it ran more time. but i dont think this has made any difference, can you tell me if this makes any sense,please see the results below .

 

PRE-PROD PDCR JOB

AMPCPUTime

TotalIOCount

PHYSIO

PHYSIOKB

18/10/2018

10,199.71

9,976,921.00

1092359

178152572

19/10/2018

25,951.27

10,654,526.00

1129626

183230792

CHANGE

15,751.56

677,605.00

37267

5078220

 

PS- The query took almost double time on 19th 

Ambassador

Re: PDCR Job took more time to copy approximately same amount of data

Hi,

 

Assuming that 19/10 is the 'long run'... that is using @2.5 times the amount of CPU compared to the previous day.

 

  • During that run was the system CPU constrained?
  • Which WD (i.e. priority) were each run using?
  • Did they both start/finish in the same WD?
  • Were there any throttle delays?

Cheers,

Dave

 

Ward Analytics Ltd - information in motion
www: http://www.ward-analytics.com