NASLite Network Attached Storage

CAT5e is just fine for gigabit speeds. I install Leviton CAT5e gigamax connectors and panels all the time and they are run at gigabit speeds day in and out with no dropped or malformed packets. FYI, CAT6 is the Gigabit spec. CAT7 is yet to really be adopted and to my knowledge Siemon is the only manufacture of product that performs at the proposed CAT7 spec. Truth be told Fibre is cheap and much better for the gigabit and beyond speeds that are comming. I recomend that fibre be pulled with CAT6 cable in both homes and businesses. The raw fibre is cheap and termination is not needed unless you are going to use it.

As far as the NIC is concerned I only use 3Com or Intel nics. The one exception being the NIC in my NAS box and that is a Netgear GA621 Gigabit Fibre card. It works just great and I use fibre to my managed switch.

Good quality cable, termination, NICs and swithces are needed when you expect the best in performance from a network. Most of the networks we speek of here are not using anything more than patch cables between the computer and the switch. Most of the switches use the Realtech switch chips which are ok but not great. The Broadcom chips are great chips though and there are switches out there that do use them. The Gateway switches used them last I checked that were being unloaded on Ebay.

Mike

Hi Tony,
yes, I have got the GT running too. Runs equal to XT and MT. No problems at all.

Have a nice day,
Roland

That's good news...

Jeez, I didn't mean to start a ****-storm here. I was just disappointed with the performance I have been able to achieve and did a simple, quick-and-dirty REAL WORLD test against another box on my small network. And now I have people nattering about pirated OS's, switch latency, NIC cables, rebooting cycles, etc.

If my results are correct, it means nothing more or less than what I said in the initial message in this post: perhaps it is time to stop adding FEATURES for a while and add some PERFORMANCE instead.

If you believe my results are not correct, post YOUR benchmark results... and tell me what I need to tweak in my NASLite2-USB installation to achieve the same results. I used PassMark Performance Test v5.0, running on a WinXP workstation and accessing the server(s) as mapped drives.

And just what new "FEATURES" have been added?

Ralph, you are still just quibbling. I am sure there is a changelog that lists all new features that have been added. You probably have it yourself.

From the website:

*Support for SATA, SCSI, USB and FireWire connected fixed disk drives
*Support for multiple IDE, SATA, SCSI, USB and FireWire interfaces
*Dependable hardware RAID support
*Remote NFS volumes re-export

Given a choice, as a matter of priority, I would have preferred a doubling of throughput for PATA drives over any of these.

Again... if you feel my results are wrong, post yours...

.

Here's the changelog, the link is available on the homepage :

http://www.serverelements.com/history.p ... Lite-2-USB


I'm not seeing the new "FEATURES" please quantify your statement.

Here's the original post I responded to, not the edited one above.




lmoseley, your edited post with supposeded new "FEATURES" is totally impertinent to what you asked for.

I don't know what you are trying to say, Ralph, since inpertinent is not a real word.

The features that I listed in my edited post are new features that have been added.

In the earlier versions, you had performance (throughput) to a point that the 10/100 network components were the limiting factor. Now, as people move to GB LAN, this is no longer the case, and so perhaps it is time to turn your attention to tweaking performance again.

And that is exactly what I said in the first post. Perhaps not as politely as I could have said it, but it remains true nonetheless.

If you feel my results are wrong, post yours...

.

huh

I think I'm confused.

People have been using Gbit with NASLite since the original floppy disk releases with pretty respectable performance results. Should I interpret your comment above as “NASLite v1 is faster than NASLite-2”?

We’ll do our part and review the currently implemented protocol optimizations for anything that may’ve been missed, but I’m pretty sure we are where we need to be. Although there are some further tweaks that can be made to the TCP/IP layer, the gains are negligible and the performance with respect to non-windows clients suffers in return.

No, you should interpret my comment to mean that NASLite+ v1 was fast enough so that the limiting factor was the 10/100 network system, and NASLite2 is also fast enough that the limiting factor was the 10/100 network system. Speed improvements made to either product would hardly have been noticed by the typical user.

But now as people move to the GB LAN, the network is no longer the limiting factor. The GB LAN can move more data than NASLite can supply. Improvements in NASLite speed will be noticeable to the GB LAN user.

A typical, reasonably modern 7200 rpm ATA-100 drive can deliver 40 MB/sec of sequential read throughput to the Windows machine it is mounted in. That same Windows system can serve out about 23 MB/sec through a GB network. NASLite2, in my test, could only serve 12 MB/sec. That is only about 30% of the raw speed of the drive... a lot of MB/sec is being left on the table. Somewhere. Somehow.

Where and how, I don't know. I do know that I expected that a lean dedicated NASLite server running on new hardware would be able to beat the performance of an old clunker machine running Win2000 Workstation. In my tests, it couldn't. That was the reason for this post.

If you believe my tests were wrong, post your benchmarks. And tell me what I need to do to match them.

.

I really don’t mean disrespect by this, but stating hardware capability figures will not help me see your point here. Let me attempt and help you understand what is my problem with your initial post:

Looking at your syslog here (and I’m assuming this is the system in question): http://www.serverelements.com/phpBB2/vi ... php?t=1207

Looking at this:

This kind of stuff in the frequency displayed in your syslog indicates a problem with communication between the drives and the IDE controller. Mind you, I have no details to elaborate on this due to lack of access to your machine, but you can see my point.

Looking at this:

I’m willing to bet that your NIC shares IRQ 11 with USB. That is also something to consider.

My point to your original post (My words):


You are not benchmarking what NASLite-2 is capable of doing, but what it can do despite the problems your hardware is posing. That’s all there is to it.

I think you pretty much told us previously your unwilling to conduct any benchmarking in a fair manner, and it's obvious by your skewed results.

I think I spelled everythig correctly this time

I installed NASLite2 in the manner described in the dox. I am not a Unix guru, as you guys no doubt are. If you think that the NIC IRQ is the problem, I'll move the card to another slot. As far as the mobo-drive communication, I don't know what to say about that. New mobo, new drives, new cables, jumpered properly, adequate power supply. POST indicates no problems.

For all of your nitpicking, Ralph and Tony, neither of you have posted your own benchmarks to show the speeds that an EXPERT, using a personally-tuned NASLite2 system, can achieve. Unless you do, I'll have to think that mine are representative results.

If you believe my tests were wrong, post your benchmarks.

.