12 Collecting and Reporting Statistics

Oracle GoldenGate provides MEASFLS and MEASRPT to assist in determining sizing and configuration parameters for TMF and non-TMF based data synchronization. MEASFLS gathers statistics using Measure. MEASRPT interprets Measure statistics and produces a report that assists in the sizing process.

This chapter includes the following sections:

12.1 Running MEASFLS

The function of MEASFLS is to start Measure processes that collect data about file I/O at specified intervals.

MEASFLS uses a parameter file to identify the list of files to measure, the reporting period, and other parameters. Create or edit the parameter file using the NonStop editor.

Run MEASFLS using the following syntax.

TACL> RUN MEASFLS /IN parameter_file/

MEASFLS initiates activity measurements. After MEASFLS finishes, wait for Measure activity to complete before running MEASRPT.

Example 12-1 MEASFLS Parameter File

--
-- MEASFLS Parameters
--
START 2010-03-07 01:00
DURATION 7 DAYS
INTERVAL 1 HOUR
MEASFILES $DATA2.MEASDAT.MDAT
PURGEMEASFILES ON
WILDCARD $*.*.*

This example parameter file:

  • Runs Measure for seven days with a reporting interval of one hour, starting at March 7, 2010 at 01:00.

  • Gathers statistics for all files on the system every hour.

  • Extracts Measure data into files named $DATA2.MEASDAT.MDATnn.

  • Purges the previous measurement using the same Measure data files.

12.2 MEASRPT Statistical Output

The MEASRPT report outputs the statistics generated with MEASFLS and Measure. These statistics are output during individual intervals as cumulative totals, identifying both peak and summary activity levels.

Many of the statistics can be displayed as either absolute numbers or on a per-second basis. See the MEASRPT parameter REPORTRATE in Reference for Oracle GoldenGate on HP NonStop Guardian.

Statistic Category Description

CPU Busy

Shows the average percent busy across processors during each period and over the total period measured. The report includes a peak utilization rate. These figures provide insight into the current utilization rates on the system and can be used to project utilization rates after Oracle GoldenGate is installed and running.

Top 10 Peak Periods

This section of the report displays the top 10 most active periods on the system as a whole, according to three criteria.

  1. The total number of output operations across all files (inserts, updates and deletes).

  2. The number of bytes generated during the period, without compression, plus 48 bytes per record. This number approximates the number of bytes that would be transferred in a given period to the target system if compression is not used. Each record transferred by Oracle GoldenGate is preceded by a 48-byte header.

  3. The number of bytes generated during the period, accounting for compression on update records, plus 48 bytes. This number is an estimate of the number of bytes that would be transferred to the target system over the period.

These numbers, especially the last two, provide an estimate of the communications bandwidth requirements.

Top 10 Peak Files

This section of the report displays the most active files over a single interval.

For non-TMF applications, distributing files across multiple Logger processes becomes a matter of figuring out the total number of I/Os a single Logger can handle, then assigning files to each Logger accordingly.

This section of the report helps identify how files might be assigned to multiple trails and audit trails, and downstream, over different communications channels.

Statistics Across All Files

This section of the report displays a summary of activity across all files for each interval of the report. These statistics point out peak I/O activity, and aid in sizing bandwidth requirements.

This section also reports the peak byte generation over the intended retention period. Use this to size total disk space requirements for an extended outage on the target system. You can alter the intended retention period with the MEASRPT RETENTION parameter.

Statistics By File

This section of the report displays statistics for each interval for every file extracted and selected for the report. You can exclude specific files from the report (with the EXCLUDEFILES parameter), even after the statistics have been gathered. Excluding files can be useful if not all data will be delivered.

Statistics for each file include cache hit rate, file busy figures, bytes generated, and the number of operations. Generally, Top 10 Peak Files provides enough information for distributing the load effectively, but occasionally more detail is desired.

You can limit statistics by file to the most active files (LISTLIMIT parameter) or omit them completely (FILEDETAILS parameter).

Statistics by Program

This section of the report displays statistics for each interval for every program that updates files specified in the report (If a program updates files excluded from the report, those statistics are not included in the report either.)

Understanding which programs are causing the most updates can be useful for spotting opportunities to eliminate certain delivery activity. For example, a FUP LOAD on one system may not be required on the other system.

You can exclude specific programs from the report (using the EXCLUDEPROGRAMS parameter). When a program is excluded from the report, file activity caused by the program is excluded from all report sections.

You can limit statistics by program to the most active programs (LISTLIMIT parameter) or omit them completely (PROGDETAILS parameter).

TMF Transaction Report

This section of the report displays statistics regarding TMF transactions during each interval and in total. This section also displays the programs that began the most transactions and programs that aborted the most transactions.


12.2.1 Running MEASRPT

MEASRPT outputs statistics gathered by MEASFLS that are pertinent to the Oracle GoldenGate sizing process.

MEASRPT uses a parameter file to identify the list of files to report about, activity reporting criteria, the reporting period, and other parameters. Create or edit the parameter file using the NonStop editor.

Run MEASRPT from TACL using the following syntax.

TACL> RUN MEASRPT /IN param_file, OUT report_file/

Example 12-2 MEASRPT Parameter File

--
-- MEASRPT Parameters
--
MEASFILES $DATA2.MEASDAT.MDAT
INCLUDEFILE $*.GGSDAT.*
GETSQL OFF
GETAUDITED OFF
LISTLIMIT 30

The example parameter file:

  • Specifies that Measure data extracted into files named $DATA2.MEASDAT.MDATnn be used for the report

  • Reports file activity for files in subvolume GGSDAT on any volume, including secondary partitions (even though more data may have been extracted)

  • Ignores (by default) alternate key files, SQL indexes, and NonStop files (codes 1-1000)

  • Explicitly omits SQL tables from the report

  • Omits audited files from the report

  • Limits detailed output to the 30 most active files and programs