Maybe this will help others as well.
I've managed to get 37gig/hr with our 155 ATM network.
08/24/1999 00:35:58,BACKUP(CLC) - Database: IS,
Type: 0,
Actual bytes: 12,329,255,868,
Secs: 1992.47,
Kb/Sec: 6187.91,
Exchange server:WD21EXCH002,
ADSM server: D:\adsm\agentexc\dsm.opt,
Status: 12,329,255,868 bytes sent in 1992.47 seconds(6187.91 Kb/sec)
Our ADSM Servers are IBM PC Server 325's with Dual PII 266's with 256meg
RAM. With 3COM PCI 155 ATM Nic's.
Our Exchange Servers are IBM Netfinity 5500's Dual PII 450s with 896meg RAM.
With ForeRunner 155 ATM Nic's.
ADSM Server Version 3.1.2.20 and 3.1.2.40 (both running on WinNT 4.0, SP5)
We have 2 exchange servers each going to a different adsm server
(found a bug in 3.1.2.40 and haven't upgraded the other server yet - but
performance is the same in regard to exchange)
I use Exchange Agent FIXTEST - Version 1.1.0.0 "A"
Installed on an ADSM client level of 3.1.0.6.
Well, here's the command lines I use on the clients:
excdsmc /backup:dir,full /adsmoptfile:%exc_dir%\dsm.opt
/logfile:%exc_dir%\excsincr.log /buffers:8,640 /adsmnode:wd21exch002_exch
/adsmpwd:wd21exch002_Exch >> d:\adsm\agentexc\excsched.log
excdsmc /backup:is,full /adsmoptfile:%exc_dir%\dsm.opt
/logfile:%exc_dir%\excsincr.log /buffers:8,640 /adsmnode:wd21exch002_exch
/adsmpwd:wd21exch002_Exch >> d:\adsm\agentexc\excsched.log
My exchange schedules for backups run a 12:01am everyday.
(SQL runs at 10:00pm, and FileSystem backups run at 08:00pm)
Note that I am backing up directly to tape (bypassing the disk pool for our
2 exchange servers)
We use 2 IBM Magstar L32-3575's each with 4 3570C drives (7meg/second native
14meg/second compressed)
DSM.OPT for Exchange:
COMMMethod TCPip
TCPPort 1500
TCPServeraddress wdadsm002
TCPBUFFSIZE 31
TCPWINDOWSIZE 63
TXNBYTELIMIT 25600
NODename wd21exch002_exch
password wd21exch002_exch
COMPRESSIon Off
PASSWORDAccess prompt
TAPEPrompt Yes
SCHEDMODE Polling
ADSM Server Stats:
10gig db (5 2gig volumes) - Mirrored In Sequential Mode
2gig log (2 1gig volumes) - Mirrored In Sequential Mode
DSMSERV.OPT
************************
COMMmethod HTTP
HTTPPort 1580
COMMmethod TCPIP
TCPPort 1500
TCPWindowsize 63
TCPNODELAY No
MSGINTerval 1
MAXSessions 50
BUFPoolsize 65536
LOGPoolsize 512
COMMTimeout 900
IDLETimeout 45
TXNGroupmax 250
DATEformat 1
TIMEformat 1
NUMberformat 1
MESsageformat 1
LANGuage AMENG
EXPInterval 168
EXPQUiet No
MIRRORRead DB Verify
MIRRORWrite DB Sequential
MIRRORRead LOG Verify
MIRRORWrite LOG Sequential
MOVEBatchsize 40
MOVESizethresh 500
STAtusmsgcnt 1
VOLUMEHistory volhist.out
DEVCONFig devcnfg.out
RESTOREINTerval 1440
USELARGebuffers Yes
DISABLESCHEDS No
EVENTSERVER Yes
REQSYS Yes
Some things to keep in mind:
Make sure that your exchange maintenance schedule isn't running at the same
time you scheduled backups for (Your exchange admin might have to work
around your backup schedule for best performance)
Pick a time at night when Exchange usage is lowest.
Try not to have migration or reclamation running on the ADSM server when
exchange is backingup.
We did have exchange going to disk pool for a while and were only getting
about 1.2meg/s (7-9gig/hr) or so. We believe it to be a problem with our
EMC boxes. But since we only have 1 exchange server per ADSM server, I
decided to dump right to tape (we have more than enough drives to handle
it). Performance skyrocketed. spiking at 6.2meg/s.
Some troubleshooting things to try:
If you are using TCP/IP as your commethod, run a trace from your exchange
box to your ADSM server
From a command prompt on the Exchange box type:
TRACERT <ADSM_SERVER_NAME_HERE>
This will list all routers an such in between Exch and ADSM, A long time ago
we were getting like 500k/s... Thought to myself that isn't right for an ATM
network, ran a trace and found a 16/4 Token ring router handling the 155 ATM
packets... Talk about bottleneck.
I admit that I am new to ADSM and could have an option wrong (maybe even
more speed?). If anyone sees anything please let me know.
-Wes
=========================================================================
|
