Topic
  • 4 replies
  • Latest Post - ‏2012-05-16T22:42:30Z by ITM61
robbie_hernandez
robbie_hernandez
9 Posts

Pinned topic How does ITM rollup services data in regards to current state

‏2012-03-21T07:02:39Z |
Hi, I was looking at the DB2 Warehouse that houses our services data. The historical data that stores the information on an hourly basis, for example, what exactly does the Pruning agent do by default when storing services data? If for instance, the service was read as running 2 times in one hour and not running twice, what should the hourly current_state read? I tried running a service and bringing it down a couple times. I don't see any reading of it being at a not_running state. If there is some documentation that explains this better or if anyone knows the answer much appreciated.
Updated on 2012-05-16T22:42:30Z at 2012-05-16T22:42:30Z by ITM61
  • Roseberry
    Roseberry
    311 Posts

    Re: How does ITM rollup services data in regards to current state

    ‏2012-03-21T11:57:57Z  
    Hi,
    Which Data Collection frequency are you using (independent of the frequency of the situations)? If one collects one reading every hour then only one reading will be available in the table (even if a situation is checking every 30 secondes or if no situation is defined). For your tests you might consider changing the frequency to every minute or so.
    And I'm also wondering what would be the average of one reading at 100% alive and one at 100% dead (computed by the Summarization part of the S&PA). Half-alive or half-dead? For that one, time must be included so the answer would be "alive 50% of the time".

    • - - - - - - Luc Roseberry
    Tivoli Certified
    ITIL Master Certified
    Roseberry International
    http://www.roseberry-international.com
  • robbie_hernandez
    robbie_hernandez
    9 Posts

    Re: How does ITM rollup services data in regards to current state

    ‏2012-03-21T19:17:05Z  
    • Roseberry
    • ‏2012-03-21T11:57:57Z
    Hi,
    Which Data Collection frequency are you using (independent of the frequency of the situations)? If one collects one reading every hour then only one reading will be available in the table (even if a situation is checking every 30 secondes or if no situation is defined). For your tests you might consider changing the frequency to every minute or so.
    And I'm also wondering what would be the average of one reading at 100% alive and one at 100% dead (computed by the Summarization part of the S&PA). Half-alive or half-dead? For that one, time must be included so the answer would be "alive 50% of the time".

    • - - - - - - Luc Roseberry
    Tivoli Certified
    ITIL Master Certified
    Roseberry International
    http://www.roseberry-international.com
    The check is happening every 15 minutes. Even when down about half the time it seems it rolled up into the hourly range as still running.
  • Chuck_Tribolet
    Chuck_Tribolet
    164 Posts

    Re: How does ITM rollup services data in regards to current state

    ‏2012-04-11T18:20:04Z  
    BTW, I believe Robbie is looking at the NT_Services attribute group.

    Chuck Tribolet
    triblet@almaden.ibm.com
    http://www.almaden.ibm.com/cs/people/triblet/
  • ITM61
    ITM61
    168 Posts

    Re: How does ITM rollup services data in regards to current state

    ‏2012-05-16T22:42:30Z  
    I assume you are speaking about collecting the attribute group NT_Services

    Hence you have following table in your warehouse:
    • TABLE NAME : NT_Services
    • TABLE CODE : NTSERVICE
    CREATE TABLE "ITMUSER"."NT_Services" (
    "TMZDIFF" INTEGER NOT NULL ,
    "WRITETIME" CHAR(16) NOT NULL ,
    "Server_Name" VARCHAR (64) NOT NULL ,
    "Timestamp" CHAR (16) ,
    "Display_Name" VARCHAR (64) ,
    "Current_State" VARCHAR (20) ,
    "Start_Type" CHAR (16) ,
    "Binary_Path" VARCHAR (64) ,
    "Account_ID" VARCHAR (32) ,
    "Load_Order_Group" VARCHAR (20) ,
    "Service_Name" VARCHAR (64) ,
    "Display_Name_U" VARCHAR (400) ,
    "Binary_Path_U" VARCHAR (392) ,
    "Account_ID_U" VARCHAR (52) ,
    "Service_Name_U" VARCHAR (256) );
    there you get the Service state for each sample.

    if you are looking now into the summarized table (e.g Hours, but the others are the same)
    • TABLE NAME : NT_Services_H
    • TABLE CODE : NTSERVICE_H
    CREATE TABLE "ITMUSER"."NT_Services_H" (
    "SAMPLES" INTEGER NOT NULL ,
    "DBTMZDIFF" INTEGER NOT NULL ,
    "DBWRITETIME" CHAR (16) NOT NULL ,
    "SHIFTPERIOD" INTEGER NOT NULL ,
    "VACATIONPERIOD" INTEGER NOT NULL ,
    "TMZDIFF" INTEGER NOT NULL ,
    "WRITETIME" CHAR (16) NOT NULL ,
    "Server_Name" VARCHAR (64) NOT NULL ,
    "Timestamp" CHAR (16) ,
    "LAT_Display_Name" VARCHAR (64) ,
    "LAT_Current_State" VARCHAR (20) ,
    "LAT_Start_Type" CHAR (16) ,
    "LAT_Binary_Path" VARCHAR (64) ,
    "LAT_Account_ID" VARCHAR (32) ,
    "LAT_Load_Order_Group" VARCHAR (20) ,
    "Service_Name" VARCHAR (64) ,
    "LAT_Display_Name_U" VARCHAR (400) ,
    "LAT_Binary_Path_U" VARCHAR (392) ,
    "LAT_Account_ID_U" VARCHAR (52) ,
    "Service_Name_U" VARCHAR (256) );

    you see that the summarization provides you only the latest value for each Service in the defind timeframe.

    this is a very nice example that it is usefull to have first a look what you get if you summerize the raw data. In this example i doubt that there is any value to summarize.
    How S&P is summarizing an attribute depends on the type of this attribute (e.g Counter, Gauge,...) this can be seen in the dock?? (e.g in your case docknt) files at TEPS

    hope this helps