External Networked Hard Drive Transfer Speeds

Hello everyone,

For those who do not need more info, why is my transfer speed at 20% of the wireless cards' maximums with an external HD connected to 1 computer?

For more info, continue reading.

I have a WD MyBook 320GB USB 2.0 Hard Drive (7200 RPM) connected to my main desktop computer. That computer has a USB 2.0 wireless adapter (54Mb/s). I have a laptop (5400 RPM) with an internal 54 Mb/s wireless card and the external hard drive is mapped as a network drive on my laptop for ease of access, sharing, storage, etc. My question is about transfer rates:

If I am doing my math right, a 261 megabyte file (~2088 megabits) should be transferring from my external HD to my laptop in about 40 seconds (2088 mbps / 54 mbps = 38 2/3 s). The reality is it takes closer to more than 4 minutes. I checked task managers inbuilt network monitor, and of the 54 mb per second, both computers show usage hovering around 19-20%.

So...why is my transfer speed at 20% of the wireless cards' maximums? Is this a HD i/o bottleneck? I hardly think it's THAT slow...again, if my math is right, I'm moving at about 10.8 mb/s, 1.3 MB/s? Wheres the bottleneck if I can connect my hard drive via USB to my laptop and get much faster transfer speeds? The same file transfers in about 10 seconds, or 208.8 mb/s (26.1MB/s) when connected through USB 2.0. So, it's 2000% faster through USB than my 54mb/s wireless connection which is actually running at about 11mb/s.

So the bottleneck is definitely occuring in the wireless stage somewhere, but I don't know where. Is it how windows handles wireless network transfers or something? Perhaps I should disable active AV guards when transferring? But still, it's moving at 26 MB/s through USB...Maybe my USB wireless adapter on the desktop is to blame. I will look into that, too, but in the meantime, if there are any other ideas, I definitely appreciate any input!


Thanks for any help you can offer!

 

Oh man, why can't we put issues in one or two lines only

 

Lol that should be a rule

OP: I dont know, I have had the same issue and cannot fix it. I even tried hard wire and that did nothing.

 

There is an overhead for any network connection, wired or wireless, that will impact file transfer speeds. In a wireless configuration, you will never see full speed - it's doubtful you'll get more than 25-30% utilization. Plus, you have two operating systems that monitor and direct traffic. Don't forget that the computer must transfer data to and from the external drive via the USB bus, which also has an overhead.

Theoretical speeds are nice, but in a networked environment (using the term loosely, since your setup is merely a workgroup and not a true networked environment) you will always experience much slower transfer speeds when compared to a simple data transfer from an external drive to a computer.

 

I'll let you guys duke it out on this; sorry about all the extra information that may or may not be helpful. In any case, thanks for the replies that were constructive...

@kegobeer:
I don't think the USB transfer speed is a bottleneck, being that 2.0 speeds are 480mb/s. Well above overhead.

Hey, maybe we'll get G speeds with N now!

So I guess you are saying the bottleneck is the result of "two operating systems that monitor and direct traffic" and "an overhead for any network connection, wired or wireless", or a combination thereof? That's a huge performance loss for a vague reason, although I am not doubtful that it's possible. I would expect to see a similar loss through wired then, too, and that idea is supported by NAS Ghosts' observation that wired connections still suffer large performance loss.

Well, thank you for the help and explanations. I'd definitely like to get to the bottom of this, so I'll keep looking around.

@Wirelessman
I've updated the original post for your ease of reading.

 

USB isn't a bottleneck, it's just one more slight slowdown that can contribute to slow transfers across a network. Any time you have multiple devices connect via a wired or wireless network, small overheads for each device can add up to a large slowdown. If you have a dedicated server running server software, with high speed drives and top of the line equipment, instead of a desktop with an external drive, your transfer speeds will increase.

802.11n is much faster than 802.11g, and since I've made the switch I've noticed a huge reduction in time when transferring files. It's nowhere near the speed of a standard 100BASE-TX, and definitely pales by comparison to 1000BASE-TX, but still a welcome improvement over 802.11g. I posted some data about 802.11n versus 100BASE-TX, and I recommend giving it a read if you have the chance.

If you haven't seen this website, check it out. Not only is it a speed calculator, it explains why you will see speed degradation.

http://www.t1shopper.com/tools/calculate/downloadcalculator.php

If you are running Vista, this website is a good read for painfully slow transfer times.

http://www.ads-links.com/index.php/how-to-fix-windows-vista-slow-network-transfer.html

 

Just for the record, that is the burst rate of USB 2.0, not the average; this is precisely why on average, Firewire 400 ( 400mb/s ) often beats USB 2.0 in overall throughput; the overall through put of USB 2.0 almost always less that half the burst rate.

 

@kilezo: Thank you, is much better. I just hate to read, I normally fall asleep

So, both the desktop and the laptop share a wireless router, and the external HD is attached to the desktop. What type of security you have in the desktop and laptop?

 

That is a fantastic point. The sustained rate of USB 2.0 is right around 33MB/s (read) and 27MB/s (write). The OP should also consider the sustained read/write of a 5400RPM drive, and all other drives incorporated in the network.

 

Actually the effective rate reaches 320Mbps for bulk transfer on a USB 2.0 hard drive when no one else is sharing the bus, what else do you have connected on the USB bus?

 

Mbps does not equal MB/s. 320Mbps (mega bits per second) is in the ballpark of 33MB/s (mega bytes per second).

 

Actually is 40MBps for bulk transfer.

 

I definitely understand the concept of layers of overhead adding up. It just seems like a LOT of overhead here, more than I can think to account for with just USB bus, wireless overhead, disk i/o speeds, and interference even. I keep thinking I should still be getting decent speeds. I tried transferring the same file directly from my desktops hard drive over the wireless to the laptop and the rate was the same, so it's not the external hard drive/wireless compatibility.

That's a very good point, and in fact I didn't know that. I wish I had firewire but I'm working with a 6 year old Dell desktop here!

Exactly. I am running Avira AntiVir Personal on both. I tried disabling the active guard on both computers and checked the CPU and running processes on both during transfer - my desktop doesn't show an AV module running but is still showing about 70% usage (although this is unaccounted for in the process list), while my laptop is at about 50% usage, with AVs scan process taking about 45%...even though I 'shut it off'. Not very straightforward there. So after that, the speeds are still the same.

Well...now that you mention it... the wireless adapter is on the same bus. Umm...is that bad?? I didn't even think about bus sharing throughput...but that's why I'm asking questions, I don't really know what's going on!

Thanks again for the replies, everyone!

 

Yes, anything sharing the bus may impact speed. Can you install an internal card?

Are you running Vista? If so, have you checked out the links I provided in my previous post?

 

Some more insight. With 11g, 54mbps after overhead is around 33mbps with Securty (WPA). Then you have 2 pc's consuming the bandwidth so divide the 33mbps by 2. so now your down to ~17mbps. any more devices also take it away. Then if you have your AV set to filter/check incoming from your local subnet that may be another hit. Now What process do you have running in background. I know most every pieces of software you install check home with the mother ship. This normally happens when you startup, but if your using IM you have that traffic in background too.

I hope this helps you in under standing what may be going on. The max I get on downloads with 11g wireless is normally around 2-3MB/s. For true testing use a FTP client software and see what you get. This will eliminate the 40% overhead you pay for running MS SMB.

 

Actually, since this is an old computer, this card is brand new USB 2.0 installed into the motherboard.

I'm sorry, I must have forgotten to mention that I am running XP on both. Pro on the laptop and Home on the desktop.

 

Definitely didn't know that 60% of the g connection was overhead. Maybe I'll set up a p2p ad hoc network for large transfers to cancel out some of the overhead. Would that work better than using a networked drive in a workgroup?

I thought since each computer has 54Mb/s throughput they wouldn't be only sharing that speed, but I guess that means that effectively they can only reach 50% of that capacity. Wouldn't the overhead, then, be shared too? Or did you factor that in already?

I tried turning off AV, and there's no other internet activity going on during this time (I can just run the network monitor for 15 minutes alone to see that there is little to no activity during that period). I have eliminated a lot of processes in the background on both my machines, and since I don't like Google I don't think many programs are sending pings or anything.

I don't know what MS SMB stands for, but I'd love ot get 2-3MB/s. Thats still over 200% faster than I'm getting now! It makes a big difference over long file transfers. Thanks for your post, it has been very informative!

 

I didn't mean the USB, I meant the wifi. Install an internal 802.11a/b/g/n card.

 

If you get an internal card, get an Intel one, the 5300 seems to be the best choice in the market nowadays.

 

SMB stands for Simple Message Block (in this context), and, as described in this MS MSDN article is

Even this very brief description suggests that the protocol includes a lot of overhead. As blue68f100 suggested, try a true file transfer protocol like ftp to see if that improves things.