Network Backups Start Fast then Gradually Slow to a Crawl.

Sorry about the long post but this has become a real problem for us, if anyone has any suggestions or are experiencing this problem, any feedback or suggestions will be greatly appreciated.

This applies to Echo Workstation (latest version) and TrueImage Home v11 (latest version). The connection is 40ft. with CAT6 shielded cables.

We have two XP x64 machines, one as a workstation and one as a backup machine connected via gigabit Ethernet through a DGL-4100 router (TCP - DHCP).

When running our weekly backup (about 280GB), the backups starts our fairly fast, (65MByte/s), but after about 30 minutes the speed is down to around 40MB/s and after an hour or so... 18MB/s and then slower and slower from there. What should be a 2 hour job turns into a 2 day job. As the network speed goes down, so do the CPU loads and network utilization. To my knowledge the backups never actually stop or stall, they just keep slowing down.

Large file transfers through Windows Explorer work fine and the same backup with another brand of backup software works fine.
Drive to drive backups also work fine.
Backing up small drives (40GB) is not a problem because they finish before the software slows down.

These machines are fully updated with the latest Acronis software, Windows updates, drivers etc.

I have tried everything I can think of:
Disable all AV/Firewall software.
Bypassed the router and ran NIC to NIC via CAT6 (static IP's).
Tried different router.
Tried a gigabit switch instead of router.
Changed NICS from onboard NICS to PCI-e Intel Pro/1000 PT NICS. (Onboard NICs are also PCI-e)
Turned off compression, set CPU affinity to one core, set backup priority to high, set Trueimageservice priority to realtime, etc.
Backing up through Network Places.
Backing up through mapped drive.
Backed up through \\SERVER\SHARE
Simple sharing on, simple sharing off.
Tried CAT5e shielded cables, tried CAT6 shielded cables.
Nothing helps.

If I set Acronis to split the backup files, it will bring it back to speed after each file split. Any interruption in the backup will bring it back to speed.

Hardware:
Workstation 1: (Using Acronis Echo Workstation)
CPU: Intel E6850 Dual Core
MB: Asus Rampage Formula - x48
MEM: 8GB DDR2800
RAID: Areca ARC-1220
HDD: 5 WD Raptors (Raid edition) in Raid 5.
NIC: (Dual) Marvell 88E8506 PCI-e gigabit.
SOUND: AuzenTech Meridian
PWR: 1200 watt NZXT & Ultra X3 1000 watt

Backup Machine:
CPU: Intel E6700 Dual Core
MB: Abit IN9 32 Max - 680i
MEM: 8GB DDR2800
RAID: Areca ARC-1231ML
HDD (Boot Set): 2 WD 250GB in Raid 0
HDD (Backup Set): 6 WD 500 GB (Raid edition) in Raid 6.
NIC: (Dual) Intel Pro/1000 PT PCI-e gigabit.
SOUND: Disabled
PWR: ToughPower 850

If anybody has any ideas, comments or suggestions on this issue, your help would be greatly appreciated.

Thank You

James

 

I had similar problems till I enabled Archive Splitting. I set it to DVD lvl and my backup perform much quicker over the Network.

The problem is how the file is being BUILT with BUFFERED I/O as the file grows in size more and more MEMORY is needed to buffer the I/O

 

Thanks Rsamples. Sounds like yet another bug in Acronis as the other brand of backup software doesn't do it. Nothing else does this.

We don't really want to to split archives. I don't see anything abnormal with memory in Process Explorer during the whole archive process. Plenty of memory left on both machines.

So my quest for an answer continues.

Thanks

 

Hello ePlanetDesign,

Thank you for choosing Acronis Remote Workstation Disk Backup Software.

Could you please collect some information to let us investigate the problem thoroughly?

Please create Windows System Information on both computers as it is described in Acronis Help Post.

Please also collect the application log of the operation:
- run the program;
- select Show log on the toolbar or from the Tools menu;
- select the corresponding log;
- click "Save the log entry to file" (the diskette icon).

Then submit a request for technical support. Attach all the collected files and information to your request along with the step-by-step description of the actions taken before the problem appears and the link to this thread. We will do our best to investigate the problem and try to provide you with a solution.

Thank you.
--
Marat Setdikov