Monday, March 18, 2013

Networking on Ambien

Are all makers of network devices pathological liars?

Here is why I ask: I recently purchased a Linksys E4200v2 450 Mbps (450 megabit per second) dual-band router and an Airlink AWLL6075 300 Mpbs usb 2.0 wireless adapter for my laptop. The expectation was to increase the read/write speed to my Iomega Home Network Drive, which advertised Ethernet speeds (10/100/1000 Mbps). (Why three speeds? Why not just say 1 Gbps max speed?)

Let's keep the math simple: 300 Mbps should give me data transmission of 30MB/s (megabytes per second), allowing for some overhead.

I was getting about 10% of that speed.

The really aggravating thing is that I was unable to get that speed regardless of the configuration:

The best results were obtained plugging the network drive directly into the Ethernet port of the laptop, which I was frankly surprised worked at all. That gave me 10MB/sec upload and 20MB/sec download as viewed from the file transfer status popup window. On the Networking tab of the Windows Task Manager, it shows ~10% of the "1Gbps" connection being used.

It turns out that these numbers are on the same order of magnitude as reviewer tests of the drive, so after noting that they don't come close to "1Gbps", I will accept them as the drive's upper limit. And in fact I get similar speeds when connecting to the drive through the router rather than directly.

But that was as good as it gets.

As soon as I switch to wireless, file transfer performance takes another hit, and it doesn't matter whether it involves the drive or even the router. These file transfers top out at 5MB/s and usually not even that. I executed transfers between four different PCs through the router. I set up an ad-hoc wireless network between computers that didn't involve the router. Nothing made a difference. I generally achieved transfer rates of 1 - 3.5 MB/s, numbers that were consistent with what I observed in Task Manager.

The folks on the other end of my tech support calls and chats took me through the obvious (change the channel; check the QoS and UnPn), the unlikely things (update the firmware) and the downright perverse (disable the firewall and anti-virus). Nothing made a difference.

The biggest disappointment was Airlink. My new "300 Mbps download / 150 Mbps" upload USB 2.0 adapter only established a link at half those rates according to both task manager and the adapter's own network monitor . . . and then proceeded to use only (IIRC) 40% or so of that bandwidth. (Note that all these computers were in the same room for testing, so link quality was not a problem.)

And to add insult to injury, the tech support people with both Airlink and Linksys said I should be happy with these speeds! Well, no, I'm not. Nobody's product is performing as advertised. Not Airlink, not Linksys, not Dell and HP.

Well, that's not quite right. When I was connecting to the Ethernet network in Afghanistan (!!! . . . but in fact, HQ provided decent network capability for itself) with my old D620, I was getting transfer speeds closer of around (IIRC, again) 45GB/s. So I'm pretty sure at least one Ethernet port has the ability to perform. But nothing else does.

On a positive note, I was able to transfer files on and off my WD Passport at 25MB/s plugged into a USB 2.0 port and faster using a USB 3.0 port. So my problem isn't the hard drive on the computer.

Any ideas?


trumwill said...

You're right, nothing will actually work at advertised speeds. I've noticed zero difference since switching from 100Mbps and 1Gbps, which is a pretty huge jump. Wireless, I've noticed some improvement switching from G to N, but I think that's more range than speed.

In my experience, the bottleneck is usually the hard drives, software, or processor of the device(s) doing the transferring. Your post gives reason to believe that's not the case, though. So I'm not sure what to say.

I'm having to repopulate my file server after a failure. So I have been moving a lot of data recently a lot of different ways. I don't know if this will help, but these are the speeds I am getting from ostensibly 1Gbps-capable transfers:

Using Computer B to transfer files from Synology to Synology NAS's: 8.5Mbps if nothing else is moving. That drops down to 2.5 or so if Computer A is running.

From Computer B to Synology (with no other traffic): 45Mbps.

From Computer C to Synology (with no other traffic or with Computer B running): 11Mbps.

Through USB: 10Mbps or less. I've very rarely ever gotten rates faster than that.

From Synology to Synology through Computer A using sync software: 4Mbps.

From Synology to Synology using Synology's software: 3Mbps.

Terms and other items:
By "Synology" I refer to Synology's NAS servers. I own two of them. The one that failed and the new one so that my data is not vulnerable in the future.

Except where otherwise noted, transfers between computer and Synology or Synology-to-Synology are through TeraCopy, a free download that transfers large files notably faster than Windows does natively (gains are lost when dealing with large numbers of small files, however).

None of this is going through the router, however. It's all going through a switch. I try to keep things off the router itself.

samsonsjawbone said...

This is the most heartrendingly first-world problem I've read about in a while.