ok.. here is what I have both hardware and software and let me state.. I'm sorry fo rth elecgth of this post... however I wanted to be certain that as much info was here for the "experts" to read.
" Servers: Clustered Exchange 2003 Enterprise **** The Cluster is in an Active Passive mode xSeries IBM 3550 2 x 1.8 Ghz Dual Core 4GB Ram Mirrored 36 GB Boot Drives (Hardware Mirror) 1GB Nic x2 QLogic iSCSI Interface
" NAS IBM DS300 NAS iSCSI interface 14 x 146 GB Note: one side of the array is assigned to the Exchange installation (approx 640GB usable) the other side is partitioned for SQL Server 2005 Std (approx 240Gb) and another 200+ unused. Each side is in a RAID5 with a hot spare
NOTE: the servers are not even breaking a sweat...
Firmware: The firmware on the servers and NAS are all up to date& The only thing left is to update the firmware on the QLogic (QLA4050C) Current rev is Driver Version: 2.1.3.28 (STOR w32) Firm Version: 2.0.0.29 ROM Version: 1.0.0.0 iSCSI Version: 0.20 Bios Version: 1.08
Network Configuration: Cisco 3560s (10/100MB) on all floors.. Core switches are 1GB& " Exchange server NIC and iSCSI connected to the core in a separate Server Room VLAN " Users VLANs are based by Floor e.g. First floor VLAN 11, 2nd Floor VLAN 12 etc.) "
Exchange config: (approx 175 users)
Exchange Data Drive (e:\) Total Size: 633GB - Free: 435 Quorum (Q:\) Total Size: 49.9GB - Free 49.9
" We have the following structure in Exchange_data on Drive E:\
NOTE: Directorates, Inactive, and SR_Management are Storage Groups&
" We also have a fairly large Public Folders
" Currently our Database is 165GB
" Our Largest user is 10.5 GB (top ten range from 5.7GB down to 3.75GB I have no way to impose restrictions now& There would be a mutiny!! NOTE: the largest user has more than 65,000 items.. the average in our org is 30,000
Ive started to make a Structure and begun to move some people out of the First Storage Group (initially our only storage Group)& and in to directorate / roll based groups (easier to manage.. and changed the way backups occur so that I can now simply restore one mailbox and not be forced to restore the entire first storage group..
" What our Users are experiencing..
1. Our largest user (10.5GB) spends minutes switching from her mailbox to her calendar, the top 10 also experience slowness. 2. All users experience the system try notifications Outlook retrieving from Exchange server and other network notifications (network warnings etc.) 3. When we (IT) create a profile on a new computer.. it takes hours and hours to download that exchange profile to the computer
What we have done&
EXMON.exe results 30 sec capture) - attached in text file
one last thing... I has been mentioned to me that I should perhaps move all logging to another LUN... what do you think?
Also.. it has been suggested that I purchase 64bit Nic Cards for the servers
ok i guess we start with ping...what are ping times like to the server? confirm link duplex settings, are they correct no mismatches? can you check network statistics to see if there are any error counts on the ports in the switch the server is plugged into. have you used a protocol analyzer to see whats going on on the wire? might need to profile I/O between the server and the storage too. if you run perfmon on the server whats the paging file doing? Are you running the /PAE switch? whats load like on the servers? potentially the problem could be disk i/o, network i/o or some other bottleneck.
to answer your questions.. Q: what are ping times like to the server A: ping time to the cluster and each server are less than 1 ms *** Q: confirm link duplex settings, are they correct no mismatches? A: Make - Broadcom BCM5708C NetExtreme II GigE NDIS VBD Client - Duplex is set to "Auto" with the following options 10 full, 10 half, 100 full, 100 half NOTE: that that connection state is actually 1GBps on Auto the rest of the settings are as follows Checksum offload "Enable" Flow Control "Auto" Large Send Offload " Enable" Locally Administered Address "Not Present" Maximum Transfer Unit "1500" Receive Side Scaling "Enable" NOTE: these settings are on bother server !!
**** Q: can you check network statistics to see if there are any error counts on the ports in the switch the server is plugged into A: we are running Cisco 3560G's on the Core Switches (both the NAS and the Exchange servers on connected here. I have verified that there are not errors. ***** Q: have you used a protocol analyzer to see whats going on on the wire? A: No... but it's most assuredly an option.... ****** Q: might need to profile I/O between the server and the storage too A: ok.. how does one do this? ******* Q: if you run perfmon on the server whats the paging file doing? A: It's huge.. I've just changed the settings. NOTE: the servers have 4GB of RAM now.. at one point they only had 2GB ******** Q: Are you running the /PAE switch? A: What is this switch for ? ********* Q: whats load like on the servers? A: They are not even breaking a sweat!!! PF load is low and CPU lower still ********** Q: potentially the problem could be disk i/o, network i/o or some other bottleneck. A: All things exchange are on the same partition. Logs etc as stated in my initial question. Should I be moving something?
Second part..
The users for the most part are not managing anything.. I have to beat them on the head to get them to clear up their inboxes, deleted Items etc.. even then... for the most part they are just creating folders under the inbox and moving items there.. therefore not affecting the overall size nor the number oif itmes in the inbox.. (ie: it remains the same)
Is there anyone out there to offer some suggestions.. !!???
NOTE: I'm not using the /PAE switch... here is my boot.ini [boot loader] timeout=30 default=multi(0)disk(0)rdisk(0)partition(1)\WINDOWS [operating systems] multi(0)disk(0)rdisk(0)partition(1)\WINDOWS="Windows Server 2003, Enterprise" /noexecute=optout /fastdetect /3GB /USERVA=3030
