Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

We have an EMC San - CX4-240C filled with 15k (4Gb) Fibre Channel disks setup in RAID 5. Our applications that run on this SAN do primarily Sequential read and write of large text data files. I'm testing throughput and seeing a high water mark around 125 MB/sec when nothing else is running on the server. That seems slow...

In contrast, we have another server attached to a Dell MD1200 SAS based system filled with 15k drives as well. We have a RAID 10 on that configuration and get about 485 MB/sec of performance.

Is a DAS based device always going to smoke a fibre channel SAN for this type of performance comparison? Doesn't 125 MB/sec seem slow for pure sequential access on this type of SAN? I know there are a number of other considerations, but looking for some general feedback.

share|improve this question
1  
How are your servers connected to the SAN? –  EEAA Sep 22 '11 at 4:24
1  
Yes, yo ubasically give no information. Yes, DAS is faster in sequential because the DAS link has more bandwidth. Who cares about the discs speeds when your server is connected with 1gigabit to the SAN? –  TomTom Sep 22 '11 at 4:33
1  
125MB/s looks remarkably like 1/8th of 1Gbps. Are you saturating a gigabit link to the SAN? –  Slartibartfast Sep 22 '11 at 4:35
    
Yes, that is a valid assumption. Which obviousl shows how clueless some people are - high end san is totally worthless if you need throughput and then connect it with a narrow little pipe. That said, 125mb/s is a LOT of random io ;) –  TomTom Sep 22 '11 at 5:43
add comment

1 Answer

Are other devices using that spindle, or shelf, or FC link to the SAN from the switch?

Your test isn't necessarily happening in isolation on the SAN, whereas it likely is with the DAS. Combine that with the fact that the disks in the DAS are probably newer and faster than those in a CX4, and that RAID 10 is faster than RAID 5, and there's certainly enough other variables that placing the blame on the fact that the SAN is a SAN isn't terribly appropriate.

So, to more directly answer your question: no; a SAN is not inherently slower than a DAS in any way.

share|improve this answer
    
Yes, there are other VM's hooked into that SAN, but I conducted the test when there was little to no activity on the SAN. So the test was as isolated as possible without shutting down the other servers connected to the SAN. The disks in the SAN are 15k Fibre Channel 300 GB disks, and we purchased them 9 months ago - right at the same time we purchased the DAS device/ disks. I would like to believe that that configuration of the SAN plays a big role in things. Any idea on what the throughput different between RAID 5 and RAID 10 will is? –  erbz Sep 22 '11 at 4:43
    
@erbz As I asked above, how are your servers connected to the SAN? –  EEAA Sep 22 '11 at 5:30
    
The san is connected through a 4Gb fiber channel switch. Having said that, don't have enough details about the configuration - I am just the guy on the outside objectively questioning if setup and performance is optimal or could be improved. –  erbz Sep 23 '11 at 18:11
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.