| 
		
		
			 
		 | 
	
| 
			 
  | 
	|
Parameter Tuning Guidelines
Table C-1 describes the result of increasing or decreasing various OpenVMS and BEA MessageQ parameters.
Table C-1 Parameter Tuning Guidelines
| 
			 Action  | 
		
			 Advantage  | 
		
			 Disadvantage  | 
		
			 Min/Max Value  | 
		
			 Parameter Affected  | 
	
| 
			 Increase message buffer count  | 
		
			 Allows higher number of pending messages.  | 
		
			 MessageQ consumes more page file and global pages, and the page count increases for each process.  | 
		
			 8-50,000 for each buffer pool  | 
		
			 GBLPAGFIL   | 
	
| 
			 Decrease message buffer count  | 
		
			 Frees up page file for other programs.  | 
		
			 Increases the probability of lost messages or poor performance due to resource exhaustion.  | 
		
			 .  | 
		
			 .  | 
	
| 
			 Increase message buffer reserve count  | 
		
			 Allocates a higher number of message buffers for exclusive MessageQ internal use. Results in higher probability of unblocking and internal messages being delivered sucessfully.  | 
		
			 Reduces the available number of message buffers for general use.  | 
		
			 0 to all buffers in pool  | 
		
			 .  | 
	
| 
			 Decrease message buffer reserve count  | 
		
			 Increases available number of buffers.  | 
		
			 Increases probability of lost messages or poor performance due to resource exhaustion.  | 
		
			 .  | 
		
			 .  | 
	
| 
			 Increase message buffer size  | 
		
			 Allows transfer of larger blocks of data in a single call.  | 
		
			 Consumes extra memory when data fills only a small portion of the buffer. Increases the size of each MessageQ process due to the creation of the local MessageQ buffers; two to four buffers depending on the features utilized.  | 
		
			 1-31998 (small)   | 
		
			 PGFLQUO   | 
	
| 
			 Decrease message buffer size  | 
		
			 Higher utilization of available memory.  | 
		
			 Increases program complexity to disassemble and reassemble the data buffers when larger data transfers are required. Some features of MessageQ might not work because of the inability to allocate a large enough buffer.  | 
		
			 .  | 
		
			 .  | 
	
| 
			 Increase receive message quota  | 
		
			 Allows a higher number of pending messages to build up during a burst condition.  | 
		
			 A queue might consume a disproportionate amount of resources on its pending queue, causing other parts of the system to have problems.  | 
		
			 0 bytes-  | 
		
			 .  | 
	
| 
			 Decrease receive message quota  | 
		
			 Limits resource consumption of slow service rate queues or queues with no reader.  | 
		
			 Queues might be unable to receive large messages. Messages might be lost or delayed because they cannot be queued.  | 
		
			 .  | 
		
			 .  | 
	
| 
			 Increase MRS NUM_DQF_AREAS  | 
		
			 Allows more unconfirmed messages to be stored before a UMA is taken.  | 
		
			 Uses more disk space and MRS Server resources.  | 
		
			 100-1000000  | 
		
			 .  | 
	
| 
			 Decrease MRS NUM_DQF_AREAS  | 
		
			 Reduces disk and MRS Server resource consumption. Reduces time to locate a given message.  | 
		
			 Reduces probability of storing a message.  | 
		
			 .  | 
		
			 .  | 
	
| 
			 Increase MRS CACHE_PERCENTAGE  | 
		
			 Allows the queue to have more outstanding messages, which results in more overlap of the MRS Server I/O when applications process recoverable messages. This quota is a percentage of the byte or message receive quota.  | 
		
			 Could cause the MRS Server to consume most or all of the receive byte/message quota on queue attachment.  | 
		
			 1-100  | 
		
			 .  | 
	
| 
			 Decrease MRS CACHE_PERCENTAGE  | 
		
			 Reduces consumption of MessageQ buffers on queue attachment.  | 
		
			 Lowers the amount of I/O overlap for reading messages out of the DQF and delivering the messages to the appropriate queue.  | 
		
			 .  | 
		
			 .  | 
	
| 
			 Increase MRS Maximum Queues  | 
		
			 Allows the MRS Server to track more local and remote queues.  | 
		
			 Causes the MRS Server to grow larger to contain these data structures.  | 
		
			 128-2147483647  | 
		
			 FILLM   | 
	
| 
			 Decrease MRS Maximum Queues  | 
		
			 Reduces the size of the MRS Server.  | 
		
			 The MRS Server might be unable to track all the queues requested.  | 
		
			 .  | 
		
			 .  | 
	
| 
			 Increase MRS Maximum Messages  | 
		
			 Allows the MRS Server to track more messages  | 
		
			 Causes the MRS Server to grow larger to contain these data structures.  | 
		
			 128-2147483647  | 
		
			 PGFLQUO   | 
	
| 
			 Decrease MRS Maximum Messages  | 
		
			 Reduces the size of the MRS Server.  | 
		
			 The MRS Server might be unable to track all the messages requested.  | 
		
			 .  | 
		
			 .  | 
	
| 
			 Increase MRS NUM_SAF_AREAS  | 
		
			 Allows the MRS Server to handle more remote queues.  | 
		
			 Could increase the number of files and disk blocks used by the MRS Server.  | 
		
			 0-1000000  | 
		
			 .  | 
	
| 
			 Increase MRS NUM_SAF_AREAS  | 
		
			 Limits the total number of remote queues that can have pending SAF messages.  | 
		
			 Might be unable to track all queues requested.  | 
		
			 .  | 
		
			 .  | 
	
| 
			 Increas MRS AREA_SIZE  | 
		
			 Reduces disk fragmentation and number of files needed to store messages.  | 
		
			 Less efficient use of disk space.  | 
		
			 128-16384 blocks for each file.  | 
		
			 FILLM  | 
	
| 
			 Decrease MRS AREA_SIZE  | 
		
			 More efficient use of disk space.  | 
		
			 Could cause an inordinate number of disk files to be created. Causes disk and directory fragmentation.  | 
		
			 .  | 
		
			 .  | 
	
| 
			 Increase MRS Redelivery Timer  | 
		
			 Less CPU overhead due to reduced retries.  | 
		
			 Less responsive to changes in available resources.  | 
		
			 0-5000  | 
		
			 .  | 
	
| 
			 Decrease MRS Redelivery Timer  | 
		
			 More responsive in systems with limited resources.  | 
		
			 Could cause an inordinate amount of CPU attempting to retry message delivery.  | 
		
			 .  | 
		
			 .  | 
	
| 
			 
  | 
	
| 
			 
			Copyright © 2000 BEA Systems, Inc. All rights reserved. 
			  |