Attachment uploads are currently disabled. This is a temporary situation and will resume as normal in the coming days.
Solved!
Go to SolutionDN
darmstrong.navi
7 Posts
0
7038
August 30th, 2010 17:00
ML6000 throughput?
What are reasonable and expected Job Rates to see in Backup Exec on an ML6000 with two LTO4 drives? Within the last two days (and this is pretty constant for the last two months), I see everywhere between 1.00MB/min (it took 5:53:25 to backup 660,896,312) to 3,280MB/min (it took 00:01:52 to backup 917,181,476). Those rates are for the same drive.
I've never used a dual drive tape library before. Does the ML6000 have a dedicated bus for each drive, or do both drives share the same bus? We have an ML6000EM waiting to go in with two more drives. What is the bus situation on that? Will it be sharing the same bus with the current two drives?
darmstrong.navi
7 Posts
571
0
Posted August 31st, 2010 10:00
Bob,
Thanks for the thorough reply. My environment is roughly as follows. The backup server is a PowerEdge R610 running Server 2003 R2 x64 and Backup Exec 2010. The ML6000 is connected to the server either via fiber channel or SAS. Windows shows it on Bus 0 with a Device Instance Id of "SCSI\CHANGER..." so I am going to deduce that it is SAS. I will check that next time I am at the data center.
The backup targets are all remote Windows servers using RAWS. We are running two SANs, an EMC CX300 and CX4/120. The LUNs are exposed to the servers and shared out through Windows. Those shares are then backed up with RAWS. The network is GigE, with no LACP or bonded connections.
Right now I have two backup jobs running. Both of them are pulling from the same cluster, same SAN. One of them is getting 1,766MB/min and the other is getting 2MB/min.
I'm wondering if the NIC is saturated. The public connection has a pretty stable utilization of 40-50%. Dell just did a SAN analysis and the SAN itself is operating very well. There are not any red flags or causes for concern as far as IOPS, latency or any of the other metrics. The analysis did include a 48 data capture window that included numerous backups (including the ones that are currently running).
I have inherited this environment. I am wondering if there isn't a better way to be protecting the data on the SAN. My understanding has always been that it is okay to use something like Backup Exec to protect the server OS and application files, but for the SAN we should be taking snap shots or using some sort of CDP product.
DELL-Bob D
899 Posts
571
1
Posted August 31st, 2010 09:00
Your speeds are slow....you should expect the LTO4 drives to run roughly 105-110MB/sec optimally w/out compression. The drive is spec'd at 120MB/sec but you will give up around 10% for ISV overhead. Keep in mind this is an optimal setup. Slower performance will result as you introduce bottlenecks.. ie. network backups, slower servers. Nothing in the ML6K will hold it back from running these speeds w/ 1 drive or 16drives simultaneously , any throughput issues will be coming upstream ie. server environment.
The drives are cabled separately, so SAS drives will be on a different bus, Fibre drives can be on a separate bus or zoned to be run off a single HBA, scsi drives can also be separate or daisy chained.
If you describe your setup for the FORUM folks may be able to offer some suggestions for you to tweak to get closer to the optimal mark.
support.dell.com has a tool ITDT you can DL to test performance. It should be listed under the ML6K product.
Bob