For those not already familiar with the Teradata Workload Analyzer (TWA), TWA is one of the products affiliated with TASM (Teradata Active System Management). It provides guidance for the database administrator in defining the workloads for Teradata through the analysis of log data collected over a period of time and by offering recommendations on the workload definitions, classification criteria, and service level goals. It also provides a migration feature from priority scheduler.

Analysis of an individual workload can be initiated by selecting the ‘Analyze Workload’ right-click option available at workload report or clicking ‘Analyze’ tree node option. To further refine the initial set of workloads into one or more additional workloads, TWA uses DBQL data for workload analysis. Theoretically, a workload can be further sub classified into multiple workloads through additional classification criterion. Encouraging the user to sub-classify on any and all possible classification criterion would lead to confusion and many unnecessary workloads. Also, considering the operational performance points, there is a cap on the number of workloads for a database.

Workload Analyzer will guide the DBA towards appropriate classification criterion. At any given point in the analysis, the user is allowed to choose correlation and distribution parameters in the drop-down list, and then click ‘Analyze’ to analyze the associated usage patterns. He can drill deeper within a chosen cluster, or re-analyze by choosing different correlation and distribution parameters. Through trial-and-error and visualization, the user will decide which parameters identify the request group he desires to isolate most effectively. This trial-and-error process is streamlined by providing the user with distinct count and distribution range insight without having to ‘Analyze’. This can eliminate fruitless visualizations on a single user, or a tight distribution, for example.

The overall flow of the GUI can be better understood by looking at the flow chart below.

Flow Chart

This drill down is a recursive process for deeper analysis on correlation and distribution parameters. If not satisfied with the current analysis parameters, the DBA can select more appropriate parameters, as guided by reviewing distinct values and range for those other parameters.
For example, with respect to the distinct value counts, one particular workload could display the following characteristics:

  • User Name (24)
  • Applications (1)
  • Account Name (1)
  • Client addresses (2)
  • Queryband (3)
  •  Function (3)
  •  Urgency (1)
  •  AggLevel (8)
  • Estimated Processing Time (0-1000 secs)
  • AMP Count (0-1)

In this example, the DBA would be empowered to know that there is only 1 distinct application or account, and they all run at the same urgency, so trying to identify a correlation against different application, account or urgency values would be a wasted effort. However, the opportunity for correlation does exist with Users, Function and AggLevel. The DBA could pursue those correlation options. Similarly for the distribution parameter ranges: An estimated processing time range from 0 to 1000 seconds suggests a large variation of requests are included in this workload. The opportunity for identifying clusters is higher than if the estimated processing time range was simply 0-1 second.

The DBA can add clusters to the current workload for deeper analysis or clusters can be split off into a new workload. DBA can repeat this process until good set of workloads is defined or all unassigned clusters are assigned to workloads.

TWA uses an assigned and unassigned cluster concept. Each cluster (Accounts, Users, QueryBands) found during analysis are initially unassigned. Selected clusters are assigned after adding clusters to the current workload for deeper analysis or after splitting out into a new workload. Unassigned clusters remain for subsequent action by the DBA if desired. TWA brings back all unassigned clusters if the same analysis parameter is clicked again after displaying the informational message below.

Message

If unassigned clusters are not acted on by the DBA, the associated requests will be relegated to a different workload once the ruleset changes are saved. For example, consider the following set of six workloads that were generated after the 1st level of analysis on Accounts, where Workload A is defined with classification Account=A:

The DBA decides he would like to analyze workload A, who consumes 35% of the CPU. Based on some criteria (e.g. client user), he determines that one element should be isolated, and treated different than the other elements. He has two choices on how to do this: either split or add classification to existing workload.

If the user splits on the particular element, the result is a new workload, A2, with classification Account=A and Client User = xyz. A2 automatically has a higher evaluation order than the original workload A to assure client users of xyz execute within A2, and all other client users will execute within A. The CPU distribution divided between the old (A) and new (A2) workload as follows:

Alternatively, if the user chose to instead add classification to existing workload, (so that the workload classification of A is now Account=A and Client User = xyz), un-chosen elements will be designated “unassigned”, as depicted in the following diagram. If not further acted upon, they will end up executing within WD-Default because no other workload exists that would capture requests with classification Account=A and NOT client user = xyz.

It is suggested, to avoid accidental relegation of unassigned clusters to WD-Default or some other unexpected WD, that drill-down probes begin their first analysis step using the Split option. Additional refinements should then be done using the “add classification to” option against that new workload, so that unassigned requests will be relegated back to the original workload. (Example 2 below will demonstrate this technique.)

The DBA can select correlation parameters (“Who” and “Where”) and distribution parameters (“What” and “Exception”) at each depth of analysis. The DBA can also review the workload by viewing classification list after each depth of analysis and perform Undo (if needed). The Undo operation can be used to undo any previous analysis performed. It will delete assigned clusters from workload classification and bring them back as unassigned clusters for new add/split operations.

A Drill-down Example:

Here is a detailed example exploring the mechanics of workload analysis functionality and all the information provided within.

Step 1:

Select “Analysis->New Workload Recommendations” option. Define the DBQL inputs and Account String category for initial level of analysis.

Select all unassigned request group to new workload, WD_ABC.
At this point, the CPU distribution of all workloads (1) is shown below:

Step 2:

Select ‘Analyze Workload’ right-click option for cluster analysis. The ‘Analyze Workload” tabbed window as shown as below is displayed to select data filters for analysis.

Step 3:

Step 4:

The ‘Current workload classification’ list will display a summarization of the classification:
The numbers adjacent to current classification criteria type will show the total number of elements for correlation and min and max value for distribution classification parameters. E.g.

  • Account String (2) => Total 2 accounts are classified.
  • Estimated Processing Time (0 – 200 secs) => Workload is classified for Estimated Processing Time between 0 to 200 secs.

The DBA can view the detailed classification and exception definitions by clicking “View Classification” and “View Exception” buttons.

Step 5:

Select appropriate Correlation and distribution parameters. Note that there are 10 distinct Applications found for WD-ABC workload, making it a good candidate to analyze deeper. The following shows that analysis visualized:

Step 6:

Note that the ‘TWA’ Application has significantly higher CPU/IO and Avg EstimatedProcTime than other applications. It is selected for deeper analysis using the “Add Application clusters for deeper analysis” option on the right-click menu. The Report and Graph is updated for remaining unassigned clusters, while ‘TWA’ is assigned on the current workload for deeper analysis.

DBA can view classification for the current workload by clicking ‘Data Filter’ tab.

The CPU distribution of all workloads now looks like the following:

Please note that if unassigned clusters are not added to current/new split workloads before saving rule set to database then all queries arrived for unassigned clusters will be executed part of default workload (WD-Default).

If DBA selects Application again then all unassigned 9 clusters are brought back for subsequent operations (1 is assigned to current workload).

The user can then assign all remaining clusters using either the Add-to or Split options. The user chooses to select all 9 unassigned clusters and split it to new workload called WD-Others.

The CPU distribution now looks as follows, with no unassigned requests.

Finally DBA saves the rule set to database for activation.

A second Drill-down Example:

This example uses analysis against multiple QueryBand parameters to help identify and isolate various request clusters, or provide additional granularity on request clusters. One initial workload consumed a vast majority of resources, and a more granular breakdown of that workload is desired. Also, long-running outliers were noted in the analysis, and a goal is to have these outliers classified into their own workload so that different workload management techniques applied.

Step1:

Initial workloads are defined based on Account String using auto-generate option.

The CPU distribution of all workloads looks as follows:

Workload WD-ADW-DS is selected for further analysis as it is consuming 92% of the total CPU.

Step 2:

Queryband names and values are loaded in ‘Analyze Workload’ window after clicking ‘Analyze’ tree node. This system found total 5 distinct queryband names. These Queryband names with distinct value count are loaded automatically in QueryBand Filter list for analysis. However, this list can be viewed only when queryband is selected for correlation parameter list.

  • QueryBand (5)
  • AggLevel (7)
  • Function (5)
  • Region (7)
  • TopTierApp (3)
  • Urgency (3)

There are several potentially good analysis candidates here as denoted by the distinct counts. The user selects QueryBand as correlation parameter and query band name of Function from queryband filter list, then clicks “perform analysis”.

Step 3:

Below are the correlation and distribution reports/Graphs. A Total of 5 queryband values are displayed for QueryBand Name=Function.


 

Step 4:

Notice a possible distinction with Function = MIN, who included queries far lengthier than any other queries. Select ‘MIN’ row, Right-click and Add ‘MIN’ queryband value and split it to new workload ‘WD_ADW_Outliers’. This step is needed to assure unassigned clusters fall back into the original ‘WD-ADW-DS’ workload classification.

Step 5:

The QueryBand Function = MIN is split to new workload called ADW_Outliers and the remaining 4 functions are unassigned, falling back to ADW-DS if no other action by the DBA is taken on them. The CPU distribution now looks as follows:

The Correlation/Distribution reports and graph are refreshed with the remaining 4 unassigned queryband values for next add/split operation.

Step 6:

Further drill-down will be performed on the newly split workload ‘WD_ADW_Outliers’.
Select ‘Analyze’ from ‘WD_ADW_Outliers’ workload tree node for further analysis.

Select TopTierApp from Queryband filter list and click ‘Perform Analysis’ button.

Step 7:

The longest running queries are all common to not just Function =MIN, but also QueryBand Name TopTierApp = BODSS. Select ‘BODSS’ TopTierApp queryband value and add it for deeper analysis on current analyzed workload ‘WD_ADW_Outliers’.

The CPU distribution now looks as follows, with the 2 unassigned toptierapps being relegated to the original WD_ADW_DS workload rather than being part of the WD_ADW_Outliers workload:

Step 8:

Analyzing further on TWA_Outliers, notice that the distinct count for all the correlation parameter shows 1 (only 1 distinct value). This means that all requests in this workload are coming from the same combination of who parameters (Accout ADW_DS and Queryband Name Function = MIN and Queryband Name TopTierApp = BODSS). Now only distribution parameters can be used for further drill-down analysis. The Estimated Process Time can be used as distribution parameter for since the Estimated Processing Time range for current workload is wide (0.00 -158.00 secs).
Select ‘None’ as correlation parameter and Estimated Processing Time as distribution parameter.

Click ‘Perform Analysis’ button to generate distribution graph.



The 8 queries lie in last bucket (10 bucket – Range 142.20 – 158.00) long queries. However, another 8 queries in first bucket (Range 0.00-15.80) are comparatively short running queries, with a wide gap shown in buckets 2-9.

Our goal from the start was to isolate the long running requests found within ADW-DS workload and apply different workload management to them. By adding the last bucket to the ADW_Outliers workload, we will achieve the necessary workload definition to then apply those different workload management techniques. The user selects ‘Add Estimated Processing Time clusters for deeper analysis’ to ‘WD_ADW_Outliers’ after highlighting the last bucket from distribution report, overriding the min estimated processing time to 30 seconds, and overrideing the max estimated processing time to 999999 (essentially unlimited).

The CPU distribution now looks as follows, with the very short running requests relegated back to the original ADW_DS workload.

In summary, the workload classifications of interest within this example are now as follows:
WD-ADW-DS remains in its original form:

WD_ADW_Outliers is created with a higher evaluation order than WD_ADW-DS as follows:

 

Discussion
Karam 25 comments Joined 07/09
30 Oct 2011

Lots of information listed here Birendra. Many thanks for sharing.
While following the given example , for STEP 1 , in TWA -> Analysis-> New Workload Recommendations , I chose DBQL and gave a start and end date but I'm getting the following error :
Selected non-aggregate values must be part of the associated group.[62218]

I checked DBQL log to find error code 3504 but I have no power on TDWM analysis queries. MY question is that why this error is coming? What am I missing ?

TDWM user is properly created with 183 objects (we have TD 12)

BKSahu 14 comments Joined 05/10
30 Oct 2011

Let me know what TWA version (patch version number) are you using? What is DBS version (i.e. full DBS version)?

sahu

BKSahu 14 comments Joined 05/10
30 Oct 2011

Hi Karam,
This is problem with TWA 12.0 with latest TD12.0 database. TD 12.0 did some changes for aggregate queries which trigger TWA query to fail with 3504 error.
We have fixed this issue in TWA 13.00.00.06 using DR 152136. Let me know if you can upgrade TWA with 13.0. Otherwise, i can provide the TWA 12.0 efix for you. However, you need TWA 13.0 for Nth-Level drill down feature.

sahu

Karam 25 comments Joined 07/09
31 Oct 2011

Thanks again Birendra.
I was able to generate Candidate Workload Report after installing TWA 13.10
Excited to try your recommendations !!

Smarak0604 15 comments Joined 08/12
01 May 2013

Hello,
 
I am new to TWA. I am using the 13.10 Version of TWA. Once, I select the Step1[Log as DBQL & Provide the Date Range & Category "Account String"], I am getting the Error:
 
"No 'Account String' Information Available with the given Input parameters. [62218]".
 
Can you help me out here. I have enabled DBQL Logging by executing the following Statement:
BEGIN QUERY LOGGING ON ALL;
 
And, I am also seeing Data being Populated in the DBC.DBQL* Tables. 
 

BKSahu 14 comments Joined 05/10
02 May 2013

This message is displayed only when TWA did not found any accounts in DBC.DBQLogTbl for given Date Range i.e. if you were logged the DBQL data say for 04/28/2013 to 04/30/2013 and provided the DBQL inputs Data range out of log range say (04/31/2013 to 05/02/2013) in TWA then you did not get any account/users etc. Please check in DBC.DBQLogtbl table if some rows were logged for same Data range you have entered in ‘Define DBQL Inputs’ dialog of TWA. Let me know if you still have same problems.  

sahu

Smarak0604 15 comments Joined 08/12
08 May 2013

Hello,
 
I checked the DBQLogTBL Table and found that I have entries from 23rd April onwards. So, Instead of using 1st April-30th April as DBQL Date Range, I used Date Range from 23rd April-30th April.
 
Even with any Selected Option (Account String, Users, Profile, Application), I am getting the Error:
 
No "<Selected Option>" information available with the given Input parameters.[62218]".
 
Does it check any Specific Tables in DBQL which might be Empty. Cause, Data is getting Populated in the DBQLogTbl Table.

BKSahu 14 comments Joined 05/10
08 May 2013

Hi,
 
There is no differance between Date Range 1st April-30th April or 23rd April--30th April if data was logged only 23rd April onward.
 
I feel some problem with your data dictionary table "sys_calendar.calendar". This table may not be populated correctly while executing Dip Script.
Please run the following query and see if all the dates from 1st April to 30th April are listed in results set (30 rows).
SELECT calendar_date FROM sys_calendar.calendar
WHERE ( calendar_date BETWEEN '2013-04-01' AND '2013-04-30' )
AND day_of_week in (1,2,3,4,5,6,7) Order By calendar_date;
 
Please run below query and see if you are getting any DBQL rows as output.
SELECT StartTime,FirstRespTime,(extract (second from FirstRespTime)  + (extract (minute from FirstRespTime) * 60.0 ) + (extract (hour from FirstRespTime) * 3600.0 ) + (86400.0 * (cast(FirstRespTime as Date)- cast(starttime as date)))) - (extract (second from starttime) + (extract (minute from starttime) * 60.0 ) + (extract (hour from starttime) * 3600.0 ))  AS ResTime,
  NumSteps,NumResultRows,TotalIOCount,TotalCPUTime ,UserName,d.CollectTimeStamp,
  d.QueryID,UserID,AcctString,AppID,LSN,ClientID ,ClientAddr,QueryBand,
  ProfileID,d.ProcID,WDID,NumOfActiveAMPs,OpEnvID ,SpoolUsage,
  HotAmp1CPU,HotAmp1IO, ZEROIFNULL(EstProcTime) as TheEstProcTime,
  CAST(1 AS INTEGER)  as QueryCount,TotalCPUTimeNorm,
CASE WHEN StatementType (NOT CS)LIKE  'SELECT%' THEN 'SELECT'
 WHEN StatementType (NOT CS)LIKE 'DML%' THEN 'DML'
 WHEN StatementType(NOT CS)LIKE 'DDL%' THEN 'DDL'
ELSE 'OTHERS'
END AS StatementType,AMPCPUTime
FROM (
SELECT StartTime,FirstRespTime,NumSteps,NumResultRows,TotalIOCount,
  (AMPCPUTime + ParserCPUTime) as TotalCPUTime,UserName,CollectTimeStamp,
  QueryID,UserID,AcctString,AppID,LSN,ClientID, ClientAddr,QueryBand,
  ProfileID,ProcID,WDID,NumOfActiveAMPs,OpEnvID,SpoolUsage,MaxAMPCPUTime as HotAmp1CPU,
  MaxAmpIO as HotAmp1IO,EstProcTime , (AMPCPUTimeNorm + ParserCPUTimeNorm) as TotalCPUTimeNorm,
  StatementGroup AS StatementType,AMPCPUTime
FROM DBC.DBQLOGTBL,sys_calendar.calendar
WHERE ( StartTime BETWEEN TIMESTAMP  '2013-04-01 00:00:00.00'
 AND TIMESTAMP '2013-04-30 23:59:59.99' ) 
 AND (calendar_date = CAST(StartTime as DATE))
 AND day_of_week in (1,2,3,4,5,6,7) ) d 
UNION
 ALL
SELECT CollectTimeStamp,
CASE when ResTime  < 60 then collecttimestamp + cast ( cast( ResTime as int) as interval second)
 when ResTime  > 59
 and ResTime  < 3600 then collecttimestamp + cast (cast( ResTime as int) /60 as interval minute) + cast ( cast( ResTime as int)  mod 60 as interval second)
 when ResTime > 3559
 and ResTime < 86400 then collecttimestamp + cast ( cast( ResTime as int) /3600 as interval hour) + cast ( ( cast( ResTime as int)  mod 3600)/60 as interval minute) +  cast ( (cast( ResTime as int)  mod 3600) mod 60 as interval second)
 when ResTime  > 86399
 and ResTime  < 2678400 then collecttimestamp + cast ( cast( ResTime as int) /86400 as interval day) + cast ( ( cast( ResTime as int) mod 86400)/3600 as interval hour) +  cast ( ((( cast( ResTime as int) mod 86400) mod 3600)/60) as interval minute) + cast ( (( ( cast( ResTime as int)  mod 86400) mod 3600)) mod 60 as interval second)
END ,COALESCE(QuerySeconds/NULLIFZERO(QueryCount), 0) as ResTime,
  NULL,NULL,TotalIOCount,(AMPCPUTime + ParserCPUTime) as TotalCPUTime ,
  dbase.DatabaseName,CollectTimeStamp, NULL,UserID,AcctString,
  AppID,NULL,ClientID ,ClientAddr, CAST(  NULL AS VARCHAR(6160)) AS QueryBand,
  ProfileID,ProcID,NULL,NULL,NULL ,NULL,NULL,NULL,0,QueryCount,
  (AMPCPUTimeNorm + ParserCPUTimeNorm)as TotalCPUTimeNorm,CAST(  NULL AS VARCHAR(128)) AS StatementType,
  AMPCPUTime 
FROM DBC.dbqlsummarytbl, dbc.dbase,sys_calendar.calendar
WHERE DBC.dbqlsummarytbl.UserID = dbase.DatabaseId
 AND CollectTimeStamp BETWEEN TIMESTAMP  '2013-04-01 00:00:00.00'
 AND TIMESTAMP '2013-04-30 23:59:59.99' 
 AND (calendar_date = CAST(CollectTimeStamp as DATE))
 AND day_of_week in (1,2,3,4,5,6,7) ;
 
 

sahu

Smarak0604 15 comments Joined 08/12
25 May 2013

Hello,
 
As requested, I executed the mentioned 02 SQLs and they didn't give any Result-Set.
 
FYI: I am using the Teradata Express 14 Edition via VMWare Player. Does it have something to do with the Result-Set ? Or, how to ensure all the Required DIPs are properly executed ?
 
Thanks In Advance.

BKSahu 14 comments Joined 05/10
27 May 2013

This problem nothing related to Virtual (TD Express) or non-virtual database. The Dip Script did not run correctly on your system. Due to this, sys_calendar.calendar table is empty and Join with this table is failed which resulting no rows. Please rerun the Dip script with “10 DIPCAL” to populate the calendar system table.   

sahu

geethareddy 145 comments Joined 10/11
26 Nov 2013

hi Sahu,

I am getting below error:
An error occured while loading the ruleset: RULESETNAME.
The error message returned is: Macro 'tdwm.TDWMSelectConfigurations' does not exist. and error code is 3824. [62218]
I raised an incident to Teradata. Meanwhile want to get your suggestions.
 
Thanks,
Geeta.

Thanks,
Geeta

BK185002 1 comment Joined 05/09
26 Nov 2013

Hi Geeta,
This is the problem with tdwm setup. TWA using macros to save the ruleset. Could you please execute the TDWM dip script to recreate TWA macros? Let me know which version of TWA and Database you are using?
Regards
Sahu

BKSahu 14 comments Joined 05/10
26 Nov 2013

Hi Geeta,
This is problem with your TDWM database setup. TWA is using some macros to save the ruleset which are missing in your TDWM database. Please rerun the TDWM dip script to creates the TWA macros. Let me know which version of TWA and database you are currently using?

sahu

geethareddy 145 comments Joined 10/11
26 Nov 2013

Appreicate your quick response.
TDM: 14.00.00.00. DBS: 14.00.04.06d.
I got this error when i choosen the option 'Analysis'-->'Existing Workload Analysis'.
And i am not getting any error if i choose 'Analysis'-->'New Workload Recommendations'. Any suggestion why i am not getting that error for 2nd option. And I got 'Unassigned requests report' with almost 21 account strings. I was surrpised to know that we do not have any Workload assigned to these many account strings. I am interpreting correctly? And what are the extreme downsides of it having those many Account strings under wd-default (apart from reporting).

Thanks,
Geeta

geethareddy 145 comments Joined 10/11
27 Nov 2013

And in my 'Unassigned requests report' i got account strings that have already been assigned to a some Workload . But it is reported by 'Unassigned requests report'. I am wondering why it is showing like that.
For ex: we have a id BatchUser which is already part of a different rules (throttle and one workload). BUt still its name is appearing in Unassinged report. Please clarify what is missing here.
And regarding my previous question i am suspecting the previous upgrade didnt run the DIPTDWM 100% successfully.

Thanks,
Geeta

BKSahu 14 comments Joined 05/10
27 Nov 2013

In Existing Workload Analysis, you not supposed to get unassigned requests. All requests are must be part of some workloads including default workload if TASM enabled. This case only arises when you on/off TASM (enabled and disable TASM), other case would be when DBQL captured some queries when TASM is disable. Please check if some queries in DBQL used for workload analysis having NULL WDID. This means some queries were running when TASM cat3 disabled. The same queries were classified part of some workload when TASM enabled but same queries running outside of workload when TASM was disabled. Is this happen in your setup when DBQL captured with TASM enabled and disabled? please confirm.

sahu

geethareddy 145 comments Joined 10/11
27 Nov 2013

To answer your questions.
Q)Please check if some queries in DBQL used for workload analysis having NULL WDID.
(Yes I checked one account string, its FinalWDID has "some name" and sometimes it is showing "Unassinged" for some of the queries.I noticed the same thing with other account strings too.)
Q)TASM cat3 disabled.
(i didnt  understand what does cat3 (I am guessing category3 that means workload definitions ) disabled means?
We have almost 37 WDs. I am giving this count from VP WLDesigner.).Please let me understand what you mean by cat3 disabled.
Q)The same queries were classified part of some workload when TASM enabled but same queries running outside of workload when TASM was disabled.
(I am sorry,i am confused here. For ex, if an accountstring assinged to one WD (assume WD-Payroll) then how come the DBQL shows UNASSINGED for WDID and why the same queries from that Payroll application will run outside of the workload rule (WD-Payroll).

Plesae let me know if you want me to contact via email to share better inputs.

Thank you again for your concern.
Geeta

Thanks,
Geeta

BKSahu 14 comments Joined 05/10
27 Nov 2013

1. FinalWDId should be some number. Should not be string. If some name than it is dbql bug.
2. Cat3 means workload definition.
3. When you deactivate the workload definition rules then queries will run part of no workload as ruleset was deactivated. So same account become unassigned but when workload definition rules activated than same account run part of some workload and TWA show assigned.

sahu

geethareddy 145 comments Joined 10/11
27 Nov 2013

May be i have made you confused with 1. On Q1.
Let me be clear on this. After i hit on 'Analysis'-->'New Workload Recommendations'. I got a list of Accountstrings under 'Unassigned requests report'. I have picked one of the account strings from that list to check what kind of WD that particular AS getting assinged. And I have noticed for some of the queries under that AS got some Workload name and for some of the queries got UNASSINGNED .
Yes you are right, may be its a bug.

And coming back to my original question, do you have any suggestions for me on handling, ' why i am getting big list of account strings in my 'Unassigned requests report' even those account strings already in cat3 WD rules.
And as a 2nd and 3rd attempt, i tired with the 'Unassigned requests report' with Users(2nd attempt)/Profiles(3rd attempt). i am getting the  same huge list. This time i got the report in terms of the users/profiles.

Thanks,
Geeta

BKSahu 14 comments Joined 05/10
27 Nov 2013

Ohh! You are using new workload recommendations. I am not sure why you are using this option when you have already ruleset active and running. 'New workload recommendations' feature can be used only if you want to create fresh ruleset. You will have only unassigned request as this feature does not consider analyzing existing workloads. You can refine your workloads only using 'existing workload analysis'.My all previous comments were applicable only for existing workload analysis feature.

sahu

geethareddy 145 comments Joined 10/11
27 Nov 2013

I was under the impression that New WL recommendation will gives me the report of unassigned list of Accntstring/users/profiles..
For ex: if i have defined ruleset on 100 users on Oct 2013. After that i have 10 more users added to the system. But i havent reviewed my ruleset. I thought with TWA i can make use of this option (new WL recommdn) and get the recommendation on newly added 10 users. I am sorry if i made a mess here with wrong interpretation.
And coming back to your suggestion on refining i was getting that error mentioned in previous post(The error message returned is: Macro 'tdwm.TDWMSelectConfigurations' does not exist. and error code is 3824. [62218]). I think we need to run the DIPTDWM again to resolve this. I hope i understood this correctly this time :)
Thank you so much for your detailed comments.

Thanks,
Geeta

BKSahu 14 comments Joined 05/10
27 Nov 2013

Very much correct. Please rerun the DIPTDWM to fix this problem. Please play with Existing workload analysis feature to define workloads for new 10 users. Save the changes in your ruleset. Please reactivate your ruleset to make these changes effected for database.

sahu

geethareddy 145 comments Joined 10/11
27 Nov 2013

Great, thank you.
I will keep you informed on how my exising WL anlysis went.

Thanks,
Geeta

geethareddy 145 comments Joined 10/11
27 Dec 2013

hi Sahu,
Currently we did not define the SLGs on any of the workloads. Is there anyway i can get help from TWA to find out which workloads require SLGs. I heard, TWA will help me on this, but I am not sure the exact process to find out this. Please share your thoughts on this.

Thanks,
Geeta

BKSahu 14 comments Joined 05/10
28 Dec 2013

Hi Geeta
This can be done using existing workload analysis. Please find the steps below.
1. Enable the DBQL with details Log option for the period you feel suitable to represent your workloads. One month historical data can be OK.
2. Invoke the ' Existing workload analysis' feature and enter the DBQL inputs based on step1 .
3. Generate the report and review the SLG Met column. In your case NA as SLG were not defined. Select Calculate SLG for all the WDs from tool menu. Let's SLG calculated for all the WD.
4. Select the SLG Graph node from each workload from left side tree window and review the same. Modify manually if not satisfied with auto SLG. The SLG calculated for 80% SLP but you can change for any SLP and recalculate SLG automatically.
G
5. If Happy then save the ruleset and reactivate again.

sahu

geethareddy 145 comments Joined 10/11
28 Dec 2013

Great. Thank you Sahu. I will try it and see.
Just wanted to ask one more question here, is there any way I can find out the ids(including user and application Ids) which are not classified in to any workload? I think any user who did not get assigned to any workload will fall in to WD-Default workload. Please correct me if i am wrong.

Thanks,
Geeta

geethareddy 145 comments Joined 10/11
03 Jun 2014

Hi Sahu,

 

I do not think so the Viewpoint 14.10 or 15 has the features of "Workload Analyzer" included in WLDesigner Or WLMoniitor Portlets.

Is there any plan to decommision the Workload Analyzer and embedd the features in to Workload Designer Portlet in future versions?

 

I am asking these because Viewpoint becoming more centralized place for WLM and trying to understand the special features of WLA compare to the WL oriented portlets from Viewpoint. 

I found WLA is more helpful in analyzing the existing Wloads and defining new SLGs

 

 

 

 

 

Thanks,
Geeta

You must sign in to leave a comment.