Mutliple Database ( one with secondary broker ) parameter setup...

[saroren]

This is our startup script for multiple database (two database).
1) The first database need secondary broker
2) The second database just needs 4gl connection

Will the below parameter work? Any input will be highly appreciated .


proserve db1 -B 653600 -c 350 -L 300000 -n 460 -Mn 101 -Ma 5 -Mpb 50 -Mf 3 -spin 50000 -I 300 -N tcp -S 8000 -ServerType 4GL
proserve db1 -m3 -Ma 4 -Mbp 50 -N tcp -S 9000 -ServerType SQL

proserve db2 -B 1036000 -c 350 -L 300000 -n 260 -Mn 50 -Ma 5 -Mf 3 -spin 50000 -I 300 -N tcp -S 7999 -aiarcdir /temp/ai/ -aiarcinterval 300


proapw db2
proapw db2
probiw db2
proaiw db2

proapw db1
probiw db1


Thanks
Saravana

 

[Rob Fitzpatrick]

This is our startup script for multiple database (two database).
1) The first database need secondary broker
2) The second database just needs 4gl connection

Will the below parameter work? Any input will be highly appreciated .

Depends on your definition of "work"

You're on at least 10.1A, but it would be good to know your exact version and platform.

My opinions:

proserve db1 -B 653600 -c 350 -L 300000 -n 460 -Mn 101 -Ma 5 -Mpb 50 -Mf 3 -spin 50000 -I 300 -N tcp -S 8000 -ServerType 4GL

• I don't know your DB block size or your DB size, so it's difficult to say whether -B is at the optimal value, but it's a decent size. At least it's not the default.
• I'll confess my ignorance about -c. Is that only for use with DataServers, or also for Progress databases?
• Your lock table is rather large; can you justify 1200 lock table entries per 4GL client?
• Your -n is a little low, given up to 250 remote 4GL clients and 200 SQL clients. Do you have any self-service clients?
• -Mn is correct, given 50 servers of each type and one secondary broker.
• Depending on how the clients perform, and whether you have the available RAM and file handles for more servers, you might want to decrease -Ma and increase -Mpb to improve performance. If you do change them, you will also have to increase -Mn accordingly.
• I believe 3 is the default for -Mf.
• Your -spin is pretty high; have you tested with lower values? You may have more CPU utilization than is necessary.
• -I 300? What is this for?
• You haven't specified -minport and -maxport. I always do that, to eliminate potential connectivity issues (and known Progress bugs), especially on a multi-broker machine.
• You haven't specified -bibufs; the default is probably too low for you.
• No AI? Does the data in this DB not have any business value?

proserve db1 -m3 -Ma 4 -Mbp 50 -N tcp -S 9000 -ServerType SQL

Again, I would specify -minport and -maxport to keep the server port ranges segregated.

proserve db2 -B 1036000 -c 350 -L 300000 -n 260 -Mn 50 -Ma 5 -Mf 3 -spin 50000 -I 300 -N tcp -S 7999 -aiarcdir /temp/ai/ -aiarcinterval 300

• Large buffer pool; that's good.
• Same comments as above apply for -C, -L, -n, -Mn, -Ma, -Mf, -spin, -I, -minport, -maxport, -bibufs, -aibufs.
• You're using AI; that's good. Five minutes is a pretty short interval. Good for minimizing data loss, but lots of files to deal with. I assume the /temp partition is on a separate disk from the database, and you are routinely moving the extents off to a DR box (along with the backups).

proapw db2
proapw db2
probiw db2
proaiw db2

• Good (probably).
• I would run a watchdog (prowdog db2).

proapw db1
probiw db1

• Is one APW enough for this database? Do you have buffers flushed at checkpoints?
• I would run a watchdog (prowdog db1).

Transaction performance will also be affected by your BI file configuration. If you're at the default cluster size of 512 KB, you should increase it. That could be a big bottleneck on a system with many users.