Speed Issues on fast network while in Syncro Cycle

[RobH]

I am new to ViceVersa and Clustering. I need help optimizing the settings for our situation.

We are having file server access slow down issues with our newly registered ViceVersa Pro 2.0.0.9 during some synchro cycles.
We will get long delays from trying to open/save/rename a file to opening a directory. I've isolated the issue to when ViceVersa is Syncronizing. We spent quite a bit of money on our hardware and software. We can not tolerate these slow downs because it does interrupt our production.

The hardware for our 4 servers are all the same.
Intel Case & Motherboard, two 3.6Ghz Xeons with 2GB ram, with 6 Western Digital Raptor SATA 74gb Raid5 drives.

Server10 & Server13 are Windows 2003 Standard and are DC's
Server11 & Server12 are Windows 2003 Enterprise setup as a cluster, and are member servers.

All 4 servers and most workstations are on Gigabit LAN, a few workstations are at 100Base LAN.
The Cluster servers also have a private network with a dedicated 3Com Baseline Gigabit switch.

The data on the Cluster is about 145gb with a very dense directory structure. There are about 30 users, and on average 15-20 users are connected at any given time.

DFS is also being used.

The network drives are actually mapped directly to Server11, and not the cluster root. We were running into incorrect file version issues. Sometimes users were getting older versions of files from Server12 or could not see a newly created file while they were mapped to the cluster root, so I mapped everyone directly to Server11.

What I've done so far was reduce the Performance/Bandwidth from 100% to 60%. To try to correct the speed/access issue. I sill had some users complain. I don't want to mess with the settings too much without getting some feedback from you.

This is the configuration:
Folders
- Source: "F:\Root\"
- Target: "\\Server12\Root\"
- Include Subfolders: Yes
- Create sources/targets that do not exist
Comparison
- Comparison Type: Size and Timestamp
Execution
- Method: Synchronization
- Tracking Database: "C:\Documents and Settings\All Users\Application Data\ViceVersa PRO 2\vvdb1.tdb"
- Log File: "C:\Documents and Settings\All Users\Application Data\ViceVersa PRO 2\log1.log" (max 1000 KB)
- No archive for deleted/replaced source files
- No archive for deleted/replaced target files
- Process Priority: Normal
- Speed: 60 Buffer Size: 32768
- Copy files even if timestamp changes after initial comparison
- Use 'Volume Shadow Copy' to copy open files for both source and target
- Copy file/folder permissions
- Copy file/folder auditing
- Copy file/folder owner

We look forward to your response! We really need the help to get this solved.
Thanks,
RobH

[TGRMN Software]

Do you think the slow down users are seeing is bacause of Viceversa running, ie does not happen while ViceVersa is not running?
I would try once without the Shadow Copy settings, to see if the shadow copy creation is involved. Then you could lower the speed or limit the bandwidth (see bandwidth settings in the profile settings).
_________________
--
TGRMN Software Support
http://www.tgrmn.com
http://www.compareandmerge.com

[RobH]

Hey there, thanks for the reply.

I will note that the problem does Not happen Every Time there is a sync. However, when end-users complain about the slow down, a sync is in process.

Our directory structure is very dense. We have many directories that reach 255 chars in length. The ACL is very dense as well. I’m wondering if this contributes to some of the speed issues while sync’ing.

We had outside help that setup the Cluster and ViceVersa. I know that the Enterprise cluster servers do not have shadow copy running on any drive. However, ViceVersa was setup to use the Volume Shadow Copy. I can’t answer why. When I check the sync logs, they do say that a shadow copy is being created… I just don’t understand that process.

I will turn off the shadow copy setting in ViceVersa, and report the results soon.

Thanks,
RobH

[RobH]

After turning off the Shadow Copy option, I haven't heard any complaints from users about slow network access during a replication cycle.

Thanks,
RobH