Hello everyone!

I have a question over MQ log and data filesystem sizing, have gone through below documentation and tried formulas there but the calculations aren't seems right for me while i was calculating filesystem size from volumetrics data - 

Calculating the size of the log

How to calculate the total amount of disk space for the recovery logs for one queue manager

For reference below are the volumetric figures- 
- persistent message = Yes  
- average message size (bytes) = 4194304
- Max message size (bytes) = 52428800
- Max message retention time (seconds) = 86400
- average message daily volume = 1500
- max message daily volume = 2000
- average message rate (tps) = 5
- max message rate (tps) = 10
- LogFilePages = 65535 (we use this )

from my calculation it comes to - 
- Primary log file required =  239
- secondary log files required = 72
- MQ Log filesystem size = 78 GB
- MQ Data filesystem size = 74 GB

I'm not sure if the calculation are correct - the calculations given in IBM documents above are based on long running transaction and bytes processed, there is no reference i see that the log size is calculated basis on expected volumetric data.

is it possible for anyone of you to share me readily available excel/formulas or the simple notes in easy write up to understand how this can be done.


Thank you!

The active log is the amount of log space that could potentially need to be replayed in the event of a crash recovery. If you configure a huge active log then you're potentially allowing for a huge restart time (albeit unlikely).The size of the active log should be determined primarily based upon the peak rate of log writing and the duration of the longest expected transaction. There is no need for the peak volume of persistent messages to exist in the active recovery log, and the "max message retention time" has NOTHING to do with sizing the active log.

If you are using linear logging then the amount of space available to MQ recovery log files needs to allow for a completed set of media images and the entire set of persistent messages written between media images, but this is NOT constrained by the size of the active log.
In the case of linear logging it is possible that the peak rate of log writing could occur at a time when media images were manually scheduled. If you are using automatic media imaging then the queue manager will try to spread the images around to avoid a sudden peak of log writing. 

IIRC Mark Whitlock wrote a blog about log sizing a couple of years ago.

Hello everyone!

I have a question over MQ log and data filesystem sizing, have gone through below documentation and tried formulas there but the calculations aren't seems right for me while i was calculating filesystem size from volumetrics data - 

Calculating the size of the log

How to calculate the total amount of disk space for the recovery logs for one queue manager

For reference below are the volumetric figures- 
- persistent message = Yes  
- average message size (bytes) = 4194304
- Max message size (bytes) = 52428800
- Max message retention time (seconds) = 86400
- average message daily volume = 1500
- max message daily volume = 2000
- average message rate (tps) = 5
- max message rate (tps) = 10
- LogFilePages = 65535 (we use this )

from my calculation it comes to - 
- Primary log file required =  239
- secondary log files required = 72
- MQ Log filesystem size = 78 GB
- MQ Data filesystem size = 74 GB

I'm not sure if the calculation are correct - the calculations given in IBM documents above are based on long running transaction and bytes processed, there is no reference i see that the log size is calculated basis on expected volumetric data.

is it possible for anyone of you to share me readily available excel/formulas or the simple notes in easy write up to understand how this can be done.


Thank you!

Thanks Andrew for response!

If i understand correctly what you mentioned "The size of the active log should be determined primarily based upon the peak rate of log writing and the duration of the longest expected transaction" 
and by following this - How large should I make my active log? - IBM Documentation

the only required metrics to consider are :
- Max message size (bytes) = 52428800
- Max message rate (tps) = 10
- longest running transaction (sec) = 10
- LogFilePages = 65535 
- Logging Type = circular 

logsize >= 2 x (N+1) x B = 2 x (10 sec +1) x (52428800 msgSize x 10 tps) = 11000 MiB

Minimum number of log files = (Required log size in bytes) / (LogFilePages * log file page size (4096)) = 11000 MiB / (65535 x 4096) = 42.97 = ~43

This calculation results to - 
- required LogPrimaryFiles = 43
- required LogSecondaryFiles = 13    (30% of PrimaryFiles)
- MQ Log filesystem size = 18 GB   ( required size is 14 GB but extended to make max disk usage capping at 80%)

- MQ Data filesystem size = (Max msg size x max message daily volume ) + (max message daily volume x 750 bytes per msg for persistent msg)
                                          = (52428800 x 2000) + (2000 X 750) = 104859100000 bytes
                                          = 98 GB
                                            **this is worst case scenario when all max size messages landed to queue for entire day


do you think these could be right calculations now or do I still need to work more on this please..


Thanks !

The active log is the amount of log space that could potentially need to be replayed in the event of a crash recovery. If you configure a huge active log then you're potentially allowing for a huge restart time (albeit unlikely).The size of the active log should be determined primarily based upon the peak rate of log writing and the duration of the longest expected transaction. There is no need for the peak volume of persistent messages to exist in the active recovery log, and the "max message retention time" has NOTHING to do with sizing the active log.

If you are using linear logging then the amount of space available to MQ recovery log files needs to allow for a completed set of media images and the entire set of persistent messages written between media images, but this is NOT constrained by the size of the active log.
In the case of linear logging it is possible that the peak rate of log writing could occur at a time when media images were manually scheduled. If you are using automatic media imaging then the queue manager will try to spread the images around to avoid a sudden peak of log writing. 

IIRC Mark Whitlock wrote a blog about log sizing a couple of years ago.

Hello everyone!

I have a question over MQ log and data filesystem sizing, have gone through below documentation and tried formulas there but the calculations aren't seems right for me while i was calculating filesystem size from volumetrics data - 

Calculating the size of the log

How to calculate the total amount of disk space for the recovery logs for one queue manager

For reference below are the volumetric figures- 
- persistent message = Yes  
- average message size (bytes) = 4194304
- Max message size (bytes) = 52428800
- Max message retention time (seconds) = 86400
- average message daily volume = 1500
- max message daily volume = 2000
- average message rate (tps) = 5
- max message rate (tps) = 10
- LogFilePages = 65535 (we use this )

from my calculation it comes to - 
- Primary log file required =  239
- secondary log files required = 72
- MQ Log filesystem size = 78 GB
- MQ Data filesystem size = 74 GB

I'm not sure if the calculation are correct - the calculations given in IBM documents above are based on long running transaction and bytes processed, there is no reference i see that the log size is calculated basis on expected volumetric data.

is it possible for anyone of you to share me readily available excel/formulas or the simple notes in easy write up to understand how this can be done.


Thank you!