The VirginMedia SuperHub lacks a QoS option when in 'Router' mode. These tests were done
with the SuperHub in 'Modem' Mode, with the DIR-615 connected as router. I used a linux
PC connected by cable to the DIR-615.
In the first test QoS is turned off completely. So there's no throttling of upload speeds
and as a result the download speeds are worst affected. In the second and third tests, QoS
was turned on and I selected 'manual' setting of maximum upload speed. Test 2 uses
4000kbs, and test 3 uses 3000kbs. This shows clearly how the maximum upload speed goes
down from test 2 to test 3, but the concurrent download speed improves. The DIR-615 also
allows an 'automatic' setting of the maximum upload speed. When I tried that it seemed to
choose a maximum speed of over 10000. In that mode the download performance was worse
than in test 2 or test 3. I have now left it on 3500 as a useful compromise. Others may
require a different setting, e.g. if used for gaming or with different numbers of domestic
users.
Three tests were done with with different QOS (Quality of Service Settings) on the router.
Each pair of displays was produced by running speedtest.net in one window, and as it
finished the download test starting speedtest.net in a another window. So the second
test was started while the first one was doing the upload test.
For ease of comparison the results are shown in pairs. The upper results come from
the run started first, with no interference from the other, except during the upload
phase.
The lower results in each pair come from speedtest being started just as the already
running speedtest comes to the end of the download test.
NB: These are not 'precision' tests since the performance will depend on exactly
when I start the second instance of speedtest. I tried to do it just as the first instance
completed its download so that the 'ping' test in the second instance would occur while
the first instance was changing from download to upload. Also depending on the relative
speeds the time during which instance 1 was uploading could vary and time during which
instance 2 was uploading could vary. So the upload could interfere with the download for
varying proportions of the download test time, making a significant difference to the
download average speed. Despite all this the differences in download times are striking,
though no claim is made about their being typical, or average effects. They were simply
the effects I obtained in a single run through this experiment, though similar to effects
obtained when trying out the experiment earlier.
The SuperHub has a serious lack insofar as it has nothing like the QoS option in 'router' mode.
I would strongly recommend anyone wanting to use it in a domestic situation with multiple
users, where one user's uploads (e.g. sending large email attachments or uploading
Youtube videos), or with demanding applications requiring two way communication, to set
the SuperHub in modem mode and attach a good quality router, with QoS turned on.
The Dlink DIR-615 is adequate for my purposes, though it dates from 2009, and does not
provide the very latest high speed wireless options, though I have found its wireless
range very good in our house. There are probably better routers available at modest cost
now.
Aaron Sloman
http://www.cs.bham.ac.uk/~axs
22 May 2013