View Single Post
  #16   Report Post  
Posted to uk.d-i-y
Rod Speed Rod Speed is offline
external usenet poster
 
Posts: 40,893
Default ISP speed adjusting praticality



"whisky-dave" wrote in message
...
On Thursday, 23 August 2018 20:49:47 UTC+1, Rod Speed wrote:
"Harry Bloomfield" wrote in message
news
I have just this week moved to fibre (well VDSL), from ADSL. On ADSL I
was
connecting at 18Mbs, speedtest at around fairly consistent 16Mbs.

I was moved in the early hours of Tuesday morning, since when speedtest
results have been all over the place - sometimes 31Mbs, sometimes
10Mbs.
My initial tests showed a steady, 30Mbs with the ISP supplied dumbed
down
router.

I bought a second identicle router, but with OpenWRT installed. I spent
much of Tuesday trying to gets it configuration settings to work,
swapping
to that and back to the ISP router, between times, until I finally got
the
OpenWRT to work - so multiple times.

Throughout, a connect speed was reported of a steady
40,000Mbs/9.000Mbs,


So the service is fine.


Sounds too spot on to be real.


Yeah, I did wonder about that and forgot to comment on it.

ping can also be important to record.


Only really matters for online gaming.

yet actual speedtests showed wild variations.


That's normal, it varys with the server its using.


Normally the server selcted is the fastest one at the time.


Nope.

It shouldn't vary too much from the same
server which is normally stated in the app.


It can do with the worst servers.

Are the wild variations part of the system settling in


Nope.

and how long might it go on for?


Forever. Speedtest is a lousy way to measure performance.


How else would you measure it ?


Download something from the ISP's web site.
Some of them have a file specifically for doing that.

I used to use the apple servers when they had combo
updates you could download but at busy times just
after they are realsed the apple servers were responsible
for slower connections rathe rthan the ISP.


Problem with those sites is that they often do
deliberately limit the thruput rate so that everyone
updating gets a decent rate with the immense
numbers who can be updating at once.

I get very consistent speedtest results with my RSP's
dedicated speedtest site, but very variable results
when doing iOS updates depending on who else
is updating at the same time after a major release.

And its important to measure it over a wired connection,
not over wifi. Wifi just adds another variable.


It depends what you are using, a wired
connection is pretty useless for a smartphone.


Yes, but it does make sense when you are testing the
performance of a new service to use a laptop or a
desktop with a wired connection to the modem/router.

Last week I tried to find out why my wi-fi connection was dropping.


My computer was getting 130mbs download, wi-fi was about 2.8Mbs
down 12.5Mbs up ping 11ms , and 4G was about 24.1Mbs down
13.0MBps up, ping 38ms, so 4G significantly faster than wi-fi at home.


Most likely due to others using that wifi channel in adjacent houses etc.

wi-fi at work today 77.2Mbps down, 100Mbps up ping 3ms


And thats why the performance of a new service shouldnt
be done over wifi, it just adds another level of variability.