Quote:
Originally Posted by APS
We have had the Hitron installed on the 350/20 service with 5 fixed IPs and had two issues so far:
2) We are getting big slowdowns in the middle of the day when careful speedtests show speeds of less than 20 Mb/s download but the upload seems fine. Tests earlier in the day show speeds closer to the 350 Mb/s although not quite as good as ccarmock. This is slower than the 50 Mb/s service we used to have. This makes me wonder if either the tunnel used for the fixed IP service is coming out onto the network in a congested place (we pop out near Bristol) or that they are running a throttle which is either setup incorrectly or is a very bad deal.
Anyone had similar problems?
|
Having almost the same issue with my Hitron 300Mb VMB service (30-50Mb average tests this evening). Just got it re-enabled tonight, so have only done evening testing so far.
Speedtest.net (the Flash version) also locates me around Bristol, as the endpoint of the GRE tunnel. Btw, I'm in Glasgow and have no problems maxing my residential connection.
So I also think it's a congested peering issue. Haven't seen anywhere close to a 300Mb speed test yet on any of the popular sites or servers and have tried quite a few on speedtest.net.
I suggest you try this CDN test:
http://cloudharmony.com/speedtest-for-cdn
What I noticed is that all but one of the CDN's report consistently low speeds for the 10MB/2 Threads column. While I don't expect CDN's to give very high speeds, my residential 50Mb connection performs much better and more consistently maxing out on that same column.
However, the one that gives 200Mb-300Mb consistently is Cachefly, so that may be routed differently. I also did some downloads that used a cachefly server and they were the same speed.
Let me know if you get anywhere with your investigation.
Have you tried getting VMB to switch you to dynamic IP? I know maybe not the ideal solution..
---------- Post added at 02:08 ---------- Previous post was at 01:52 ----------
Quote:
Originally Posted by ccarmock
I have 16 downstreams - which I think is where the Puma 6 based devices start to show some latency issues on the reports such as BQM, but that beyond 16 it gets worse.
Now I do know a firmware update was recently rolled out to the Hitrons - so maybe that included Intel's fix
|
From my tests it's obvious the latest firmware does include Intel's fix.
20 downstream channels:
Spikes on the BQM above are me doing speed tests and download tests. Very nice low level latency without spikes otherwise and most importantly for me, no bufferbloat!
---------- Post added at 03:30 ---------- Previous post was at 02:08 ----------
Quote:
Originally Posted by rtho782
Can anyone do a speedtest on thinkbroadband? I wonder if higher priority means better single threaded speeds.
|
I find it highly unlikely that there is any kind of "high priority" on the VM business service, in spite of sales guys patter.
You do end up on a different port and line card at the CMTS than you would on residential, which in some areas is probably less utilised than the residential line cards. But if you're in an area, node and line card that is already under-utilised you're not going to see a difference.
As for the rest of the network and peering, it's clear from many customers reports that speeds are still variable for many. Including mine on VMB (residential is fine).
Between your hub and the CMTS you're still sharing the same channels on a node that residential do. So still essentially limited in bandwidth as residential are. The downstream speeds could be 'tweaked' by the CMTS scheduler, but only to a point.
They could give you slightly better upstream priority using service flows rather than UGS, but that won't help if your area is over-utilised or at times when peak-load is reached:
https://volpefirm.com/docsis_qos/