Expected Network Speeds?

[tracerbullet]

I recently replaced my wireless setup and did some benchmarking, and wondered how I was doing. But I don't want to just give out numbers, I was hoping to instead get some info on what normal expectations should be. I've seen a lot of threads where people gave numbers and other said it was about right, but those numbers seem to be all over the place. Then of course there's bit / byte conversion, something about 2-way communication halving the speed, some %age loss for "overhead", and so on. I've searched here and Google, maybe using the wrong terms, and just can't seem to find a good answer for what one might expect.

The question (finally, sorry) is - what sort of "real life" transfer rates should one expect to see across say a wired gigabit network, a wireless G network, wireless N at "300", the same at "600", or the newer AC standard?

 

[ch33zw1z]

Wired: Depends on switch/cables/disk drives/OS. I get about 75MBps on my Gb network at home

Wireless G: I get ~2MBps you doing good

Wireless N: I get ~15MBps, IIRC...been a while since I speed tested the N at home...

Wireless is susceptible to interference from walls, power, phones, blah blah blah....so your mileage WILL vary.

Use iperf to test from machine to machine over all the mediums...

 

[tracerbullet]

Thanks, good info. Yeah, I'm assuming whatever is being tested makes sure that the network is the weakest link in the chain (ie testing using fast machines, if checking file transfer speeds then also using SSD drives, etc.). I learned quickly not to bench with my NAS, it's was pretty slow.

I'll look at iperf, I downloaded it but used the CrystalDisk instead to check speeds between equipment (mapped drives and such as needed). Good point that distance and interference change things.

Do you know how the calculations work? I know that 300mbps becomes (/8) 37.5Mb/s but what happens after? Does that go right to 50% (round off to 19Mb/s, not far from your 15 mentioned) because everything is 2-way, or is it overhead, or is it all an effect of distance, or...? Probably some of each, wondering for sure in a "typical" office or home situation.

 

[ch33zw1z]

Thanks, good info. Yeah, I'm assuming whatever is being tested makes sure that the network is the weakest link in the chain (ie testing using fast machines, if checking file transfer speeds then also using SSD drives, etc.). I learned quickly not to bench with my NAS, it's was pretty slow.

I'll look at iperf, I downloaded it but used the CrystalDisk instead to check speeds between equipment (mapped drives and such as needed). Good point that distance and interference change things.

Do you know how the calculations work? I know that 300mbps becomes (/8) 37.5Mb/s but what happens after? Does that go right to 50% (round off to 19Mb/s, not far from your 15 mentioned) because everything is 2-way, or is it overhead, or is it all an effect of distance, or...? Probably some of each, wondering for sure in a "typical" office or home situation.

There is no "for sure" in wireless. Some of each for sure though 😛. Wireless can be a picky technology. I use an Linksys E3000, no external antennae's so router orientation matters more. You'll just have to fiddle with it and see if it changes from the initial tests.

 

[tracerbullet]

There is no "for sure" in wireless. Some of each for sure though 😛. Wireless can be a picky technology. I use an Linksys E3000, no external antennae's so router orientation matters more. You'll just have to fiddle with it and see if it changes from the initial tests.

Right, I get that and am doing that bit by bit.

I guess I'm just looking for a baseline of sorts. Let's say a Wireless N, rated for 300mbps, after a number of known things are factored in could reasonably be expected to bench at 15MB/s (just randomly choosing a number). Given that, anyone measuring 15 knows they are in the ballpark. Anyone getting 20 knows they've done really damned well. Anyone getting 5 knows something might be wrong and can start looking for solutions.

The 75, 2 and 15 mentioned in your first post is exactly the kind of info I'm hoping to see, so, thanks. Hoping also someone else might share their #'s, bonus if I can wrap my head around why those #'s come about.

 

[ch33zw1z]

Right, I get that and am doing that bit by bit.

I guess I'm just looking for a baseline of sorts. Let's say a Wireless N, rated for 300mbps, after a number of known things are factored in could reasonably be expected to bench at 15MB/s (just randomly choosing a number). Given that, anyone measuring 15 knows they are in the ballpark. Anyone getting 20 knows they've done really damned well. Anyone getting 5 knows something might be wrong and can start looking for solutions.

The 75, 2 and 15 mentioned in your first post is exactly the kind of info I'm hoping to see, so, thanks. Hoping also someone else might share their #'s, bonus if I can wrap my head around why those #'s come about.

From my examples:

Gbit: 75MBps * 8 = 600Mbps (big B means Byte, little b means bit....1Byte = 8bits)

out of 1000Mbps, 600Mbps isn't too shabby. something like 30% of max bandwidth is Ethernet protocol overhead. Chances are it would be my disc drives slowing me down.

N: 15MBps * 8 = 120Mbps

Out of 300Mbps, 120Mbps maybe isn't ideal, but 15MBps is what my sustained speeds are roughly when transferring files. I don't rely on the wireless for anything but web surfing and non-bandwidth intensive tasks, so I really don't care to mess with it. Keep in mind the 30% overhead applies here as well.

G: 2MBps * 8 = 16Mbps

same as N, I don't rely on it for anything but web surfing and such so I don't care to mess with it.

 

[BrightCandle]

I see on gigabit - about 100MB/s

Wireless G (5Ghz) - 3.25MB/s - the laptop can see the router, its within 5 metres.

My understanding is with wireless is it depends on the spec. G for example is designed to peak to 54Mbit/s but in practice sustains about 22 mbit/s. N is speced to go up to 600mbit/s but you will only see half of that for any one device. Generally the overhead is at least 25% on most of the wireless spectrum, so divide by 10 the mbit/s to get the MBs estimate. But with N (300) make sure you halve it first.

ie N 300 you would expect about 300 / 2 / 10 = 15MB/s on an ideal connection.