• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

"Enterprise" drives versus normal ones

I

InorganicMatter

[H]F Junkie
Joined
Oct 19, 2004
Messages
15,461
Right now, I've got a six of these on a hardware-level-caching RAID5 controller. Serving roughly 50 people at once. It's slow. Horribly slow. Would upgrading to six of these give me a significant boost in performance, or just mediocre? Do all the advanced RAID features of those high-dollar "enterprise" drives help at all in a high-usage RAID5 scenario?
 
that's technically not an enterprise drive either. It's still SATA. The whole "Enterprise SATA" thing was an invention of the WD marketing dept.

First off what RAID card are you running? Have you tried ruinning an IO meter on the RAID to see what kind of numbers you are pushing from it? How are the users connecting to it? 10/100? Gigabit ethernet? Single port or multiple w/ load balancing?
 
It's a server and amost constantly hammered, so no IO meter. I'm going purely on user feedback (or more specifically, complaining). Users connect at 10/100, there are three 1Gb pipes going in, load balancing enabled.
 
What's it doing? File storage or database or something else? My first candidate for a quick slowness fix would be your SMB server, whether it's Windows or Samba.

Running IOMeter on the machine (even while it's serving other data) will give some idea of what kind of disk performance it's getting. Could you run it on a weekend, or after-hours, to see if you're getting decent throughput?

How much of the network is getting used? That should be an easy statistic to gather, and if you're not collecting it now, no time like the present. MRTG works on windows or *nix.

Last, what controller is it? Dell Perc something?
 
InorganicMatter said:
It's a server and amost constantly hammered, so no IO meter. I'm going purely on user feedback (or more specifically, complaining). Users connect at 10/100, there are three 1Gb pipes going in, load balancing enabled.
Click to expand...

There's your problem right there. Max speed over a 10/100 is just over 12 mbs and that is not including TCP/IP overhead. Real world, they may be getting 7 or less. I would guess the network is the bottleneck, but running an IO meter on the storage would help to confirm that.
 
unhappy_mage said:
What's it doing? File storage or database or something else? My first candidate for a quick slowness fix would be your SMB server, whether it's Windows or Samba. *snip* Last, what controller is it? Dell Perc something?
Click to expand...
This is a file server, on an Adaptec 2810SA.

So you really think that I'm bottlenecked at my network speed?
 
Adaptec could well be the problem. I'm somewhat less than wildly enthusiastic about the IOP303 processor they're using - in short, it sucks.

I'd guess network isn't the problem, at least in terms of the network itself. The protocol, though, is pretty slow in general. Check out this thread to see someone who had a similar problem on Linux and resolved it by tuning his Samba parameters. My guess is that your network is fast enough, but the protocol is inducing a lot of latency, so it's "feeling" slow to the users.

You might try a different protocol (http? scp? ftp?) and see what kind of actual performance you're getting from the machine. Then try a file copy on a user's machine (pull a 1k file from the server to the local disk and time the transfer, for example) and see if it's a latency issue or a bandwidth issue.

6 80gb disks in raid 5 is only 420gb. I'd move to two or three of the 400s that you linked in raid 1 if more capacity isn't a priority. Raid 5 is probably what's slowing down your transfer rates, if they're an issue at all.
 
It sounds like the network to me also but what size files are they retrieving? Also those drives are not true "enterprise" drives. In a multi-disk raid setup there may be a bit too much vibration, etc and the drives could possibly be struggling with seek errors, etc. Maybe not though - you really need to see what the network utilization is and how much sustained data transfer that array is getting to be sure though. I'd also check that the RAID set is not running in limp mode (failed drive). On know on the IBM SCSI RAID controllers I have on the systems up here I have the IBM RAID manager software running as a system service so I get notified immediatley if anything comes up that is degrading the arrays in any way and yes also look at your protocols as you might be able to tweak performance there as well.
 
"Users are complaining the server is slow, but they need to use the server so I cannot run IO Meter or Network Monitoring. What do you think is wrong? Disk or network latency?"

Would be like ...

"The owner is complaining that the car shudders, but he needs transportation so I cannot open the hood or perform an alignment check. What do you think is wrong? Engine balance or warped brake rotors?"

As an IT professional, I can understand your concern about running diagnostic programs while still serving customers. But we cannot be realistically expected to contribute solid advice without more background information. Anything could be the issue here ... hell it could be solar flares for all we know. :p

Cheers!
 
You must log in or register to reply here.
Back
Top