Account Strings - AG

Database

Account Strings - AG

All,

My system has the SCHMON settings (default) show below:

-----------------

Id  Partition Name    Weight  Limit

 0  Default             100    none

                        Performance Groups (0 - 249)

  Id Group Name   RP Type  Milestones & Allocation Groups[0-7]

   0 L            0  S          0.00   1

   1 M            0  Q        200.00   2      0.00   1

   2 H            0  Q         10.00   3    200.00   2      0.00   1

   3 R            0  Q         10.00   4    200.00   2      0.00   1

                        Allocation Groups (0 - 199)

 Id Type Wght  Lim  Id Type Wght  Lim  Id Type Wght  Lim  Id Type Wght  Lim

  1  N      5 none   2  N     10 none   3  N     20 none   4  N     40 none

--------------


All of the account strings on the system are defined like 

$L0+XXX

$M0+YYY

$H0+ZZZ

As the account string has an additional '0' which is not defined in the PG will the users be mapped to the appropriate AG (while logging in) or will start in the default 'M' as  the priority scheduler doesnt recognize the PGs in the account string? 

Also, how can I check which userid starts in which AG and ended in which AG, after the fact? ResUsageSPS doesn't provide me the info at the user level. Thanks.

5 REPLIES
N/A

Re: Account Strings - AG

The strings you showed should map correctly to L/M/H. Unless there's a second $ within the accountstring only the first character will be used for matching.

If you're using TASM the dbc.QryLogExceptionsV view returns a WDID and NewWDID indicating if a query was re-classified and moved into a different workload.

Dieter

Re: Account Strings - AG

Thanks Dieter. 

As we are on Appliance box we dont have full blown TASM (just filters and throttles; no workloads). In the QryLogExceptions, I just see the exceptions for the filters that we set. Wondering where can I check for the demotion path of the queries? I see that appliance has few basic query milestones set and hence wanted to check how good it is performing at the user level. Thanks.

Re: Account Strings - AG

Hi Dieter,

Is there any limit on number account strings that need to be created for one system. Currently we are each one for individual user and databse (except, for DBA's Developer's, and some system oriented id's).

For ex: If we have a application, we create 7 different objects, like AppID(Application id for that application), UserID(user id for that app), ID1(View DB ID), ID2(Logical View DB ID),ID3(MartTabl DB ID)/ (I used some random example), But like this our list of ASE are increased to 3 digit number and reaching to 4 digits shortly. I am concerned 99% of them are mapped to MEDIUM PG, and i know we need to tweak them to allocate some of the low level app IDs to LOW priority PG. And important ones to HIGH (but not to RUSH). Just wanted to know are there any best practies need to be followed to keep the number low for number of accnt strings and better way to map the existing and future account strings to L/M/H PGs(or may be create more PGs for better maintenance).

Thanks,

Geeta.

N/A

Re: Account Strings - AG

Hi Geeta,

i don't know if there's a limit, but i don't think so. 

Hopefully there's a pattern, so you can easily map account strings to PG based on wildcards in TDWM.

Re: Account Strings - AG

Thanks for quick confirmation.

May be i will check with  masters and Teradata.