Just wondering what the impact of increasing this number might be. We do quite a bit of cross-CV SQL communication using a NODE configuration. Right now, we have this set to BLOCKSIZE 8192. Could we (theoretically!) improve performance (reduce I/O) by making this a bit larger? Say, 16,384? Looks like multiples of 4,096 are most desirable.
Anyone? CA guru maybe? Would love to get a better feel for this. Also, the documentation could use some fleshing out. For example, what is the maximum number that BLOCKSIZE can be? Minimum?
This is all that appears in the SYSGEN GUIDE:
BLOcksize blk-size Specifies the size of the packet used to pass data between nodes.
Blk-size is the blocksize of the data packet sent between CVs.
The default blk-size is 8192.