Start a Conversation

Unsolved

This post is more than 5 years old

3176

November 27th, 2008 06:00

Network copy of large files to CX3-20 stops and starts

Hi,

before I contact Dell gold Support team in UK. I thought I would try out the new forum. I am a SAN notive although my SAN has been running for 1 1/2 years.

Here's my problem....

Every week we back up across the network (host - net - share - host - fibre - san) our VM's files (6 x 5gb VHDs) to the partiton on SAN. Worked smoothly for 1 1/2 years and takes about half an hour.

We physically moved SAN to new office.

Now we are seeing this behaviour...

Copying starts, we see 500mbits traffic on hosts, power path i/o per sec 180, 3 minutes pass, then net traffic goes to zero, but i/o per sec 180 continues, regular small net traffic spikes, then >5 minutes later traffic back to 500mbits. Repeat. The copy now takes hours and holds up application releases.

I can repeat this every time from virtual server multiple hosts.

My thought was SAN cache filling up (rough calc 3minutes = 8gbytes RAM) and partition performing really badly. The partition is used for archiving 300k files and 15min DR backups of things like SQL DB and logs (400mb used out of 600mb).

Would you recommend a defrag? Windows 2003 defrag seems to lock up which worries me but then the partition is large. Is there any risk other than disk performance hit?

Thanks

Gary

 

3 Posts

December 3rd, 2008 03:00

 

 

OK I seemed to have fixed it. During the office move we recabled the SPS power cables wrong. After fixing this we are getting great sustained performance.

The SAN knows the power cables are not right?! Some kind of fequency based sensor?

It would be nice to know why a fully powerd SAN drops performance...

1 Rookie

 • 

1.2K Posts

December 29th, 2008 04:00

The SPS's provides 2 main functions, battery for the cache and in the event of a complete power failiure it keeps the SP's and 1st shelf of disks powered for 2 mins, enough time to flush the cache to disk as raw data. When the arrays comes back up, this data is written back to the cache and then written to the correct lun(s). It's a method of minimising data loss.

SPA is powered from SPSA and the sense cable provides a link between the 2 (same for the B side). If it's cabled wrong, the SP doesn't have the functions of the SPS and the read/write cache gets disabled. This causes the performance issues. 

No Events found!

Top