Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[GigabitEthernet]

If I run an ethernet cable to each room from the switch in the cupboard and then have a switch in each room, am I right in saying that each device won't actually be able to transfer data at gigabit speeds because the gigabit cable bandwidth will be shared between all the devices connected to a particular switch?

[burakkucat]

You should be looking for a reel of Cat5e UTP, solid core, copper cable (and not stranded patch cable).

Here is a quick list, from Amazon. Many other sources are available.

[Ronski]

If I run an ethernet cable to each room from the switch in the cupboard and then have a switch in each room, am I right in saying that each device won't actually be able to transfer data at gigabit speeds because the gigabit cable bandwidth will be shared between all the devices connected to a particular switch?

That is correct, but you need to take in to consideration what those devices are and what they will be doing.  It's unlikely they will all need full speed all of the time.

[sevenlayermuddle]

If I run an ethernet cable to each room from the switch in the cupboard and then have a switch in each room, am I right in saying that each device won't actually be able to transfer data at gigabit speeds because the gigabit cable bandwidth will be shared between all the devices connected to a particular switch?

That is correct, but you need to take in to consideration what those devices are and what they will be doing.  It's unlikely they will all need full speed all of the time.

Hmm, I would question whether Alec's assumption is actually correct.   

Ethernet, at least over UTP, is one of the few technologies where you can actually achieve what it says on the tin.  Unlike wireless, and unlike the headline transfer rates for a lot of HDD technologies, where the advertised speeds require a large pinch of salt.

The fundamental UTP ethernet bottleneck is the cable, which for Gigabit can send 1Gbps in each direction.   But the switch needs to have somewhere to send all the data it receives, so it must be able to send that 1Gbps out on another port.  If it couldn't do that, it wouldn't be providing gigabit speeds.

Moreover, there is no fundament reason (other than the switch's own CPU performance limitations) that the switch couldn't be receiving 1Gbps on half of its ports, and sending it all back out again on the other half.  And it gets even better, because it can be receive and send 1Gbps per second at the same time on each cable, so its actually a total of 2Gbps per port.

Personally I'd suspect the switches would have some internal bottlenecks that make such 'flat out' speeds unattainable in practice.  However, a quick random search for inspiration yielded the following specs for Netgear's cheap n cheerful GS605/608 family…

http://www.netgear.co.uk/home/products/switches-and-access-points/unmanaged-switches/gs608.aspx#twoBandwidth:
under 'performance'

- GS605: 10 Gbps (non-blocking)
- GS608: 16 Gbps (non-blocking)

So they clearly feel able to claim that, in some conditions at least, the 5 port switch can handle 2Gbps (i.e. duplex) on all of its 5 ports concurrently, same for the 8 port.  Packet rate will be another bottleneck and Netgear to helpfully quote that too (click on the link), but it is less meaningful as we don't really know how any particular application will fragment the data, all into nice big ethernet frames, or into lots of tiny ones.

Having said all that, I generally expect to see at least 30% utilisation when I'm sending stuff around the house over gigabit, but I have rarely seen more than about 50-60%.   I suspect the upper limit  is mainly attributable underlying disk IO transfer bottlenecks, and the host PC drivers.

[sevenlayermuddle]

A further tip, even though Alec has not asked for it….

Some ethernet NICs (integrated or not) provide hardware support for ethernet cable test.  If you are lucky enough to have such a NIC in your system, and lucky enough that your OS provides access, they can make a difference.

I have no idea about where to find it on Mac or Linux but on my windows XP system (5 year old Dell something or other) if you find 'device manager', then open the adapter, click 'advanced' 'link speed'  and 'diagnostics' you get all the options.   They allow you to get estimates of cable 'quality', verdicts of cable tests, and even a (sometimes reasonably adequate) estimate of the distance to any cable fault. 

Proved very useful when I was doing my own wiring.

[Ronski]

@sevenlayermuddle, the way I read AlecR's question was that if he had say four PC's hung off a switch then a single link back to another switch, then depending on how the network is used that single link could restrict the ultimate link speed of each PC - if they we're all trying to access say a server on the end of that single link rather than each other.

In reality in a home network it won't cause any problems whatsoever, especially if you are only sharing a NAS or internet connection. In my setup I only have a single 1Gbps link from my server to my main 16 port switch. My office then has a 5 port switch  fed fromt the 16 port switch via a single 1Gbps connection, which feeds 3 PC's and a printer. The only time I ever notice any slow downs are when I'm shifting multi gig files to/from the server, but these normally move at around 60 to 70MB/s speeds. Testing the connection between the server and my main PC I can obtain speed results close to the maximum of 1Gbps connection.

[sevenlayermuddle]

@sevenlayermuddle, the way I read AlecR's question was that if he had say four PC's hung off a switch then a single link back to another switch, then depending on how the network is used that single link could restrict the ultimate link speed of each PC - if they we're all trying to access say a server on the end of that single link rather than each other.

Ah I see, and with hindsight I think your interpretation was probably correct. 

[GigabitEthernet]

@sevenlayermuddle, the way I read AlecR's question was that if he had say four PC's hung off a switch then a single link back to another switch, then depending on how the network is used that single link could restrict the ultimate link speed of each PC - if they we're all trying to access say a server on the end of that single link rather than each other.

In reality in a home network it won't cause any problems whatsoever, especially if you are only sharing a NAS or internet connection. In my setup I only have a single 1Gbps link from my server to my main 16 port switch. My office then has a 5 port switch  fed fromt the 16 port switch via a single 1Gbps connection, which feeds 3 PC's and a printer. The only time I ever notice any slow downs are when I'm shifting multi gig files to/from the server, but these normally move at around 60 to 70MB/s speeds. Testing the connection between the server and my main PC I can obtain speed results close to the maximum of 1Gbps connection.

You are indeed correct .