Is it possible to get 10 second or more granular interval data into daily capacity planning UDR data files? Is 1 minute data the most granular data available/configurable in an ad hoc UDR capacity planning run data collection file? Is there any way to access the Investigate History UDR data files outside of Investigate?
What is the most granular data that can be collected and shared via the Perform product? Can Capacity Planning UDR data have a sample rate of 10 seconds? What alternatives are available for publishing data at a 10 second interval? This document was originally published as resolution SLN000000227888.
|
Legacy ID:KA309024 The lowest granularity available for the UDR Capacity Planning data files used by Analyze is 1 minutes.
The Perform Collector gathers data at a 10 second sample rate and data at that sample rate is available via Investigate History which can be accessed by the Investigate GUI, Perceive (real-time data source), or PrintUDR (run against the remote agent). By default 10 second samples of data in Investigate History are only saved for 1 hour and then the data points are summarized (5 minutes for 8 hours, then 15 minutes for 40 hours). It would be possible to configure the Perform Agent to retain 10 second data for a longer period - it would just require additional disk space on the remote nodes. Outside of Investigate History there isn't a good way to capture and store 10 second samples for the long term. Generally, low granularity samples are expired from Investigate History quickly. One option would be to run PrintUDR at a fixed interval and then take the data from PrintUDR and store that in a database. But, that would require a significant amount of custom scripting. If the end users could be given access to the Investigate console (possibly through their own install of the Windows console) they might be able to see the charts they need. BMC don't license the number of console installs for the BMC Performance Assurance product, instead BMC license the number of remote agent installs so users could be provided a copy of the console to install on their PC to access the data. Another option would be to use the PrintUDR command to access the Investigate History data on the remote nodes. The PrintUDR command can be run against a local UDR data repository (either normal UDR capacity planning data for use in Analyze or a HISTORY repository on the local machine) or it can access the remote nodes over the network to read history through the Perform Agent (similar to an Investigate Drill Down). To access data over the network through the Perform Agent one would specify the '-n [remote host]' or '-N [policy file based node list]' flags without specifying the '-r [repository]' flag. This tells PrintUDR to attempt to get data directly from the remote node. It is possible to access the Investigate History UDR data files directly, but that would generally mean running the 'printudr' command on each remote node. To access Investigate History data locally on a remote node you would need to specify the -r [history repository] where [history repository] exactly matched the contents of the $BEST1_HOME/bgs/mgroup/Repository.loc file. For example, the following command will output 10 second samples directly from the Investigate History UDR data repository for 3:00 PM to 3:30 PM: > $BEST1_HOME/bgs/bin/printudr -n topgun -r /home2/perform_7.3.00/perform/history -p /home3/paska/plc/printudr_cpu.plc -b 2006/08/29/15:00 -e 2006/08/29/15:30 -i 0.10 The following similar command will output 10 second samples from Investigate History via the Perform Agent for 3:00 PM to 3:30 PM: $BEST1_HOME/bgs/bin/printudr -n topgun -p /home3/paska/plc/printudr_cpu.plc -b 2006/08/29/15:00 -e 2006/08/29/15:30 -i 0.10 The first command would need to be run on host 'topgun' (or data would need to have been copied to the node where it was run using some sort of external program). The second command can be run on any machine that has network access to topgun. In theory, it would be possible using an external method to transfer the UDR data from the remote nodes to a central location and then run PrintUDR against them (since you could put them in the /[data repository]/[hostname]/noInstance/HISTORY directory in a central location). It you were only going to run PrintUDR against a single group (such as 'System Statistics' or 'Cpu Statistcs') it would be possible to just copy those groups to the central repository and then run PrintUDR against the centralized copy of the history repository for those groups. The biggest problem is that you don't really want to be copying Investigate History files while the Perform Agent is actively writing data to those files. For 'System Statistics' and 'Cpu Statistics' which are small groups which can be copied quickly it should be possible to get them copied without corruption due to the files being modified during the copy. For a bigger group like 'Process Statistics' copying the group to a central repository while an agent was still writing to them would probably result in corrupt copies of the files on the console in at least some copy attempts. There is nothing within the Perform product itself that would handle the transfer of the Investigate History UDR data files from the remote nodes to a central machine. Related Products:
|