Cat 5 to Cat6 no diff in speed!

[kjacko]

Think i'm having a brain f*rt and can't see the wood for the trees.

My house is cat5e terminated with some cat6 cabling in place. One one leg of cat6 cable i changed the module from cat5e to cat6, terminated it in the loft into a new patch 6 panel and then a short cat5e (which shouldn't matter at all) into the 3com gigabit switch.
Basically, i just replaced cat5e components with cat6.
However, i'm getting no better speed whatsoever, and i can't figure out why.
With cat5e i'm getting transfer speeds across the network of up to 13MB/s (theoretical max for 100Mbit)
With cat6 i'm getting similar, certainly no more.
100mbit = approx. 12.5MB/s
1000mbit = approx.125MB/s

There are no power cables etc interferring, and i know that the cable leg probably doesn't conform to all the fussy cat6 criteria. But still it should give some improvement surely.
I'm not trained in category data cabling, but it's part of my job so i'm no stranger to installing network, terminating etc.... and indeed have never had any issues before like this.
I just can't get my head around it, so any obvious things that i may have missed please let me know.

Thanks
Jacko

 

[CortinaV8]

I would start looking towards hardware bus speeds.

Also what model switch is it? is it dumb of managed? is Jumbo frames enabled?

 

[kjacko]

Well....lets see;

gigabit nic in pc
cat6 patch lead from pc to cat6 wallplate
cat6 cable from wallplate to loft
terminated in cat6 patch panel
50cm cat5e patch cable to 3com gigabit switch
Cat5e patch cable from gigabit nas into gigabit switch
So far, so good afaik.

However, cos i'm transferring data to the gigabit nas, i think i'll reboot the nas and switch, and replace the 2 remaining cat5e cables in the whole process to cat6.....then we'll see.

Thanks for the input, even though i've looked more towards the hardware than its respective speeds.

 

[rebuke]

In most cases, you don't see a noticeable difference between cat5e and cat6, as gigabit (1000Base-T) was designed for use with Cat5, while 1000Base-TX was designed for Cat6, it never really took off, since 1000Base-T did a good enough job. Note that some products incorrectly advertise themselves as 1000Base-TX, even though they're not...

The other thing is, what is your NAS, as I suspect you might be hitting some limitations there, even if it has a gigabit link, it might not be able to write at that speed (e.g. I had a Linksys thing, which had two gigabit ports that it could bond, but I could never actually get >100Mbit utilisation on them, as the disk controller wasn't fast enough...)

 

[kjacko]

ooh, good point rebuke. In fact, i think i remember reading that it could transfer data up to 15MB/s.
Nads!

Maybe pc to pc over the network will prove a more accurate test?

 

[CortinaV8]

Aye, as I suspected.

We've had gigabit at work for years, pretty much since it came out. Only recently have I actually bought the hardware to make decent use out of the network.

 

[TicklyT]

I've seen quite a few cases where speed negotiation between devices fails, and links end up running at the LOWEST speed they supported. The usual fix was to fix the speed of one end of the link.

 

[kjacko]

CortinaV8, what kit did you replace? A switch with heavyier throughput?
Obviously as a home user i'm not gonna bother wth real expensive kit, but i'd be interested to know what you did to make it better.

Maybe putting a pc in with gigabit nic and home server would be better. Should one expect the controller in the pc to be better than that of the (cheapish) nas?

 

[HorseHead]

If you can, get a couple of computers with gig network cards, connect them to your switch and use the following app:

http://iperf.sourceforge.net/

It allows you to create UDP or TCP streams between the two computers to measure the bandwidth and throughput your network is capable of. It’s a great way of figuring out if the bottleneck is the read/write speed of a HDD or your NAS.

I had a 4 drive Infrant ReadyNas that was configured in RAID5 and had terrible transfer rates, regardless of the fact it had a gig port on the back. I wasn’t concerned with data safety (nothing I couldn’t put back on there) and went with a striped array (RAID 0). After it was all setup, it flew and transfer rates shot through the roof.

There are plenty of guides for iPerf, it’s a very handy tool!

 

[CortinaV8]

If the NAS was cheap then yeah, I would imagine it contains a pretty low spec controller and slow drive/s. You would probably have been better building a decent machine running a RAID of some description.

Our issues were all down to drive speeds, I inherited a load of decent spec desktop machines and a couple of servers/arrays of 1994 vintage, says it all really.

 

[HorseHead]

Infrant gear is anything but cheap. I guess you pay for the people that stand in the factory and polish them or something

 

[rebuke]

I've seen quite a few cases where speed negotiation between devices fails, and links end up running at the LOWEST speed they supported. The usual fix was to fix the speed of one end of the link.

Gigabit (1000base-T) has to auto-negotiate, you can't manually specify like you used to be able to with 10 or 100. Having said that, I suspect some devices can be set so they will either auto-negotiate gigabit or nothing...

 

[kjacko]

Yeah, my nas has two sata 500gb Spinpoints in em using raid 0, so they are actually pretty good as far as sata drives go.

Thanks for the link HorseHead, will give it a go.

 

[Click-Sure]

Personally I've never found much use for 1000Base-T (or TX for that matter) for anything other than uplinks between switches and short-run backbones.

 

[Lectrician]

You can see the link speed by hovering the mouse over the LAN icon in the system tray.

Is your gigabit switch capable of gigabit on every port? Some of the larger switches only give a few gigabit ports.