What's new

Test for speed/bottleneck in wireless to USB-connected HDD

  • SNBForums Code of Conduct

    SNBForums is a community for everyone, no matter what their level of experience.

    Please be tolerant and patient of others, especially newcomers. We are all here to share and learn!

    The rules are simple: Be patient, be nice, be helpful or be gone!

ngoonee

Occasional Visitor
I've got a Huawei WS880 which I've flashed with Asuswrt-Merlin via the Vortex fork.

A USB HDD is plugged in with a single ext3 partition, and I've got entware-ng set up, just to install rsync (so far).

Running an rsync from my linux seems to top out at 2.2 MB/s, which seems slightly slow to me. How do I check whether the bottleneck is:-

1. HDD speed (very unlikely)
2. Router
3. Wireless (N 2.4GHz connection)
4. Wireless chip on my laptop

Thanks for the help
 
Probably consider flashing back to the Huawei firmware...

AsusWRT is not supported on non-Asus HW.
 
Probably consider flashing back to the Huawei firmware...

AsusWRT is not supported on non-Asus HW.

That wouldn't solve anything as the Huawei firmware wouldn't even be able to do the role I describe. In any case this firmware works fine on the router (which is basically identical to AC68U hardware wise).
 
That wouldn't solve anything as the Huawei firmware wouldn't even be able to do the role I describe. In any case this firmware works fine on the router (which is basically identical to AC68U hardware wise).

But it's not Asus hardware is it - maybe contact XVortex - not an asus issue...
 
I'm not reporting an issue, I'm asking for advise in locating a bottleneck. As far as I can tell such advise depends only on the firmware running, and this router is running asuswrt even if its not manufactured by Asus.
 
I think ext4 is recommended over ext3?
 
I think ext4 is recommended over ext3?
If you're purely on Linux sure, but I'm willing to give up the safety net of extended journaling for the convenience of ext3 which is easier for Windows computers to mount/read directly (in case of having to recover a backup). Beats having to copy all the files out using LAN or wifi.
 
I'm not reporting an issue, I'm asking for advise in locating a bottleneck. As far as I can tell such advise depends only on the firmware running, and this router is running asuswrt even if its not manufactured by Asus.

Go talk to the XVortex folks - simply put - AsusWRT is for Asus branded/built gear...

If you have a Huawei hardware issue - that's a Huawei problem, I can't imagine that even any insight here would be useful with a AsusWRT fork on someone else's hardware...
 
If you're purely on Linux sure, but I'm willing to give up the safety net of extended journaling for the convenience of ext3 which is easier for Windows computers to mount/read directly (in case of having to recover a backup). Beats having to copy all the files out using LAN or wifi.

The routers do run on Linux. :) I think there is a few posts where RMerlin suggests as much (to use ext4, but with journaling disabled).

I agree with sfx2000 though about your issue. This does not seem like the place where you are likely to find a solution (unless XVortex visits us here).
 
I'm not sure why you've reached the conclusion I'm bringing up a hardware issue. I'm looking for advise on how to test throughput and speed. For example I just came across iperf, which seems to give me slightly over 30 Mb/s over wireless (presumably without any I/O involved), significantly better than with my rsync but still not anywhere near capacity for N networks. What else can I use to test the various components, software wise?
 
I'm not sure who you meant to address, but it may very well be a hardware issue if the Asus hardware based RMerlin firmware hasn't been modified enough to account for the specific hardware in the router you're using it on.

Let's go back to the basics. You are wirelessly backing up via RSync to a USB attached HDD?

What WiFi card does your laptop have? How far are you from the router? How big is the file (or files) that you are backing up? Have you made an attempt to find the best channel for your use?

With an 'N' Class WiFi card (and possibly not optimized channel or other interference), many small files and backing up to the USB attached drive, I don't see the performance being much better than the 2.2MB/s you stated.
 
My laptop's built-in wifi is a Qualcomm Atheros AR9485 (n-capable), and I'm backing up an assortment of files. The 2.2MB/s is measured over the largest files (at least 10 MB up to a couple of GB), I'm aware that transferring small files will be slower.

I've picked channel 6 (2.4GHz as this wifi card doesn't support 5GHz) as both channel 1 and channel 11 have neighbouring routers running. In fact some odd router is setup on channel 3 as well, but that wasn't the case (AFAICR) yesterday, I'm only seeing that today.

My iperf numbers show about 19 Mbits/sec at best (which matches the 2.2MB/s rsync gives me). Interestingly enough my raspberry pi 3 (I think it uses a combo SMSC9512/9514 ethernet/wifi adapter) gives slightly better speeds, with about 24+ Mbits/sec, while my android smartphone gets something lower (about 15 Mbits/sec).

Is this sufficient for me to conclude that the bottleneck is wifi? Anything other than iperf that I can run to test this?
 
My laptop's built-in wifi is a Qualcomm Atheros AR9485 (n-capable), and I'm backing up an assortment of files. The 2.2MB/s is measured over the largest files (at least 10 MB up to a couple of GB), I'm aware that transferring small files will be slower.

If I recall - it's nominally 802.11n, but it's a single stream, single radio card, and as such, it's not going to be very fast... (N72/N150 with wide channels)
 
If I recall - it's nominally 802.11n, but it's a single stream, single radio card, and as such, it's not going to be very fast... (N72/N150 with wide channels)
Thanks, that makes a lot of sense, and would explain why the (much less capable) Raspberry Pi gets slightly better performance. Looks like I need to look for a device with a better chip to test if that's the case. Or simply use the LAN cable...
 
My laptop's built-in wifi is a Qualcomm Atheros AR9485 (n-capable), and I'm backing up an assortment of files. The 2.2MB/s is measured over the largest files (at least 10 MB up to a couple of GB), I'm aware that transferring small files will be slower.

I've picked channel 6 (2.4GHz as this wifi card doesn't support 5GHz) as both channel 1 and channel 11 have neighbouring routers running. In fact some odd router is setup on channel 3 as well, but that wasn't the case (AFAICR) yesterday, I'm only seeing that today.

My iperf numbers show about 19 Mbits/sec at best (which matches the 2.2MB/s rsync gives me). Interestingly enough my raspberry pi 3 (I think it uses a combo SMSC9512/9514 ethernet/wifi adapter) gives slightly better speeds, with about 24+ Mbits/sec, while my android smartphone gets something lower (about 15 Mbits/sec).

Is this sufficient for me to conclude that the bottleneck is wifi? Anything other than iperf that I can run to test this?

As an fyi, 10MB is not considered a large file, but anything over about 500MB or so is. ;)

Have you actually tested channels 1 and 11 in your setup? Just because there are routers parked there doesn't mean they are using any bandwidth.

If that rogue router on channel 3 is permanent, I would bet that channel 11 will be the best channel to use in your environment, all else being equal.

Handheld devices are optimized for long battery life, not high WiFi performance (in other words, disregard it's 'performance'). The Rpi is an unknown to me, do you have anything else to use as a baseline (plugged in laptop, or wired or WiFi adaptor connected desktop)?

You also did not state the distance and obstacles between the router and your laptop.

With the overhead that WiFi itself brings to the table (and 'N' Class has higher overhead than 'AC' Class WiFi), the fact that there is a router running on channel 3 (which your router sees as interference and not another router it can coordinate itself with) and the USB overhead that increases exponentially with smaller file sizes, 2.2MB/s is looking pretty good, all things considered.
 
As an fyi, 10MB is not considered a large file, but anything over about 500MB or so is. ;)

Have you actually tested channels 1 and 11 in your setup? Just because there are routers parked there doesn't mean they are using any bandwidth.

If that rogue router on channel 3 is permanent, I would bet that channel 11 will be the best channel to use in your environment, all else being equal.

Handheld devices are optimized for long battery life, not high WiFi performance (in other words, disregard it's 'performance'). The Rpi is an unknown to me, do you have anything else to use as a baseline (plugged in laptop, or wired or WiFi adaptor connected desktop)?

You also did not state the distance and obstacles between the router and your laptop.

With the overhead that WiFi itself brings to the table (and 'N' Class has higher overhead than 'AC' Class WiFi), the fact that there is a router running on channel 3 (which your router sees as interference and not another router it can coordinate itself with) and the USB overhead that increases exponentially with smaller file sizes, 2.2MB/s is looking pretty good, all things considered.

Yeah 10MB is just my minimum because its large enough to actually show a reasonable estimate of speed. There's much larger files there.

I did try channel 11, no difference in speed. The thing is this would be time-dependent as well, I tend to only have time to check these things at 2-3 am when the kids are asleep, which isn't exactly representative of the real load during the working day.

I've compared the performance with this same laptop on LAN (which gets almost exactly 10x better speed, as expected).

Distance is roughly 22 or 23 feet I'd estimate. Only real obstacle is a 'corner', its actually in the same hall but the hall is odd shaped.

Messing around these past few months seems to indicate that this is pretty much best case for my situation. Will have to think about borrowing/trying something with a better wifi card. Or getting an AC card, since the router supports it (a first for me).
 
ngoonee, if the same laptop can achieve 22MB/s, wired, then it seems like your WiFi environment is what it is. :(

You may want to test different antennae orientations on the router and different physical placements too. Is it placed at least 10 feet above ground (earth) level? Is it at least 3 feet away from surrounding walls, ceiling and other obstacles (it's not in a cabinet, right?)?

Sometimes, moving it just a few inches or a couple of feet may make a dramatic increase in throughput.

Or, it may still be some basic incompatibility between the RMerlin firmware which is intended for a different hardware platform too.

Does using the laptop in the same room (don't put it closer than about 6 feet to the router though) make a difference? If it does, that should rule out the hardware (WiFi) and firmware compatibility questions.
 

Similar threads

Latest threads

Sign Up For SNBForums Daily Digest

Get an update of what's new every day delivered to your mailbox. Sign up here!
Top