Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Announcements

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. If you'd prefer to explore, try our test area to get started. And see here for current known issues.

New Member

ICM logger and Historical Database Server retention policy

Where do you change the retention policy for these databases? I want to maximize the amount of historical data that can be retrieved via WebView.

5 REPLIES
Blue

Re: ICM logger and Historical Database Server retention policy

Go to your HDS server, Start > Run > type "ICMDBA" > drill down on your HDS server to the instancename_hds database > on the top toolbar click "Data" > Estimate.

You can make changes to the data the HDS server stores and hit "Ok".

New Member

Re: ICM logger and Historical Database Server retention policy

OK I did that but it still only has about 2 weeks available for recall. I looked at the databases on the loggers and they are set for 2 weeks only. Does the settings need to be changed on all servers or just the HDS? Or is there a setting that I am missing in the HDS to ignore the loggers retention settings?

New Member

Re: ICM logger and Historical Database Server retention policy

For ICM 4.6.2 and previous versions change this key in the HDS. For ICM 5.0 substitute Cisco Systems for Geotel and ICM for ICR.

HKEY_LOCAL_MACHINE\SOFTWARE\GeoTel\ICR\\Distributor\RealTimeDistributor\CurrentVersion\Recovery\CurrentVersion\Purge\Retain

Under the retain key are a load of keys for each of the table types and under these are the actual tables.

These keys are dynamic so you dont have to restart the AW. (Check the Recovery process window to see the changes) These values are days (in Hex)you can increase them safely but you have to be careful if you reduce them by too larger amounts as the purge process will 'freak' out and may cause problems.

Good housekeeping ,for us, means we retain 15-35 days of data on the logger(to keep the logger overheads low) and as much data as we have diskspace on the AW HDS(Webview). Good practices for data retention should mean the your HDS should run at about 75-80% full. Use the database sizing tool as a guideline to how much diskspace you will need to keep x amount of days for eack table

New Member

Re: ICM logger and Historical Database Server retention policy

This task will have to be done individually on each of the loggers and any HDSs that exist. This doesn't really answer your question though.

ICM/IPCC v5.0 requires that Webview reporting option be installed on top of a Distributor AW that has the HDS option enabled. ALL _historical_ reporting queried by users to any HDS server will stay local and use the ‘instance’_HDS Db as its source. Under no circumstances will a Webview historical report query use the logger for its source. Modifying the HDS retention values will impact how much data exists for the Webview query to run with. Obviously this gets complicated with more than one HDS/Webview server (i.e. retention periods being equal, replication feed sources – sideA or sideB, etc.).

ICM/IPCC v5.0 default retention values are 14 days for the Logger Dbs (‘instance’_SideX) and 365 days for HDS Dbs (‘instance’_HDS Dbs). The suggestion is to be extremely cautious in modifying the Logger Db retention periods as the logger has other jobs in life – don’t cut his leg off with a monster size Db.

You mention that you made a change but there is still only 2 weeks of data (which is matching the Logger retention periods). Any chance that you’re using a version of ICM/IPCC prior to v5.0? Also, just to cover the obvious, any changes to HDS retention to a greater value only starts from that point forward. An already running HDS won’t go back in time even if the data exists on the logger. For example, HDS has 30 days of retention and the logger has 100 – you change the HDS to 100. Tomorrow the HDS will have 31 days in the Db, the next day – 32 and so on. It won’t go grab the ‘missing’ 70 past days from the logger.

Prior to ICM/IPCC v5.0, one could install Webview without the requirement of installing it onto an HDS; it just had to be a Distributor AW. This confuses things.

Say for example that HDS1 (Distributor AW w/ Webview and HDS options) is pointed to LoggerA for the replication feed and AW1 (Distributor AW w/ Webview option) is pointed to LoggerB for its feed (obviously no replication). The same query run against HDS1 (which is using the local ‘instance’_HDS Db for the report) and AW1 (which is using LoggerB’s ‘instance’_SideB Db for the report) will most certainly have different [typically slight] results since their sources are from different roots.

I find that most customers are satisfied with 15 day retention on the Loggers (this way we know that they can run a 2 week report if they had to – say via ODBC if there was a major HDS failure). Then set a 65 day retention on all HDSs (again, easily covers 2 full months). It all depends upon the volume of data being collected – an ICM/IPCC user that has 5,000 calls a day will have a much different database signature verses one with 1,000,000 calls per day (which is a discussion for a different post). Leave the loggers alone and increase the HDSs to your needs (watch your available space though and mind the 80%).

/Jeff

Jeffrey.marshall@us.didata.com

P.S. Retention is just registry keys, this doc explains the map: http://www.cisco.com/en/US/partner/products/sw/custcosw/ps1001/products_tech_note09186a008014f8b9.shtml

Blue

Re: ICM logger and Historical Database Server retention policy

dzam...also, make sure your Database Allocation size is set high enough as well. I have seen IPCC HDS servers built, then in ICMDBA when they create the database they choose 100MB for the DB Size. They then go through the DB Estimate tool. When you run the ICMDBA Estimate tool what does is the "Allocated" number in the bottom left hand corner?

/andy - berbee

1399
Views
0
Helpful
5
Replies