nmon Consolidator
 |
reads in nmon or topasout files from several AIX/Linux machines (nodes) to produce a consolidated set of data in the form of an Excel spreadsheet (requires Excel 2002 or later). |
Please note:
- The Consolidator is not officially supported by IBM and no warrantee is given or implied. Use it at your own risk.
- No support is offered by IBM. Please post any questions you have to the NMON forum (see link below)
- nmon Consolidator is the personal project of the author - Stephen Atkins.
Introduction
A separate sheet is generated for each major performance statistic and graphs are automatically generated showing summary data for each server and for the installation as a whole. The tool also allows nodes to be grouped together and will automatically calculate group totals and group averages for each statistic. Because the graphs are pre-defined the user is free to edit the titles, colours and fonts to suit their own requirements and can simply delete unwanted charts to reduce the amount of output.
Administrators who tend partitioned servers will find this tool provides the ability to get an overview of an entire machine "at-a-glance and provides the opportunity for modelling different partitioning scenarios (e.g. moving dedicated partitions into the shared pool).
The ability to view performance data for all nodes side by side will be of particular benefit to administrators of clustered applications (e.g. Oracle RAC) and allows the easy identification of load balancing issues.
The other major use of this tool will be for modelling server consolidation scenarios. The tool provides the ability to specify a performance metric (e.g. rPerf or SPECint value) for each server and for this to be used to assess total capacity requirements for a group. As grouping is dynamic, different scenarios can be assessed without the need to re-process the raw data.
A unique feature of the tool are charts showing achieved or potential savings due to virtualisation.
Features in V1.5:
- The Wavg and Max values on the data sheets are now the true values
- 3D charts are now used for the Wavg charts, this is required to support the above
- There is a new sheet for Trend_Graphs - this is to assist in capacity planning
- ToD_Graphs can now be line graphs if required and will have dates in the legend if "Trend" is selected
- A new sheet is provided for Relative I/O Content analysis
- A recovery process is now invoked for file with >65K lines - this will usually work if the files have CRLF line-endings
- Improved handling of files with varying numbers of intervals
- Some parameters have been relocated to the "Sheets" file
- The PRINTER parameter is now dynamic and applies to the "PrintAll" buttons
Instructions for use
Make sure you have the latest version of NMON
- Use NMON to capture data to a file making sure that each file has no more than 250 intervals. I suggest that you collect 96 intervals of 15-minutes each - this can easily be done using the -X option of NMON. The Consolidator will not process files with >65K lines so I strongly recommend that you do not collect TOP data.
- FTP the files to your PC (using the ASCII or TEXT options)
- Now open the nmon_consolidator spreadsheet, click on the "Analyse nmon data" button and select the .nmon files to be processed.
Version 1.6 beta
Adds the following features:
- New facility for averaging interval data - useful for files with more than 250 samples
- Removes the need for files to have CRLF line endings
- New option for excluding FScache memory usage for AIX partitions
Links
Example Output
These graphs are generated from the example data that you download above.
| Time of day graphs for the whole system showing CPU capacity Utilisation, Disk KB/sec, Network KB/sec etc. |
 |
| Time of day graphs for each node or user-defined group showing CPU capacity Utilisation, Disk KB/sec, Network KB/sec etc. |
 |
| Trend graphs for use in capacity planning exercises (designed for use with multiple files from the same LPAR). |
 |
| Graphs showing Avg/Wavg/Max figures for each statistic for all nodes in the system (useful for monitoring load imbalances in cluster configurations) |
 |
| Graphs showing potential or achieved savings due to virtualisation |
 |
