Highspeed ISP testing >100mbps

Markverhyden

Markverhyden

Well-Known Member
Reaction score
11,005
Location
Raleigh, NC
I'll have to test a new circuit from the modem/core router for many sites. So I was wondering if anyone has had any problems when they are spec'd over 100mbps. Many times I can just use one of many testing sites, whether they are the ISP or things like speedtest.net, etc.

The last few have not behaved properly and we ended up using iperf. Not sure why but the d/l is way less than the u/l and both are way off. The most recent was a VZ fiber, at the Schlitz place I mentioned. ONT in the basement, single fiber connection to an Adtran 4660. Two copper ports active with two public IP blocks, both provisioned for 300mbps. Using the traditional methods mentioned above I had d/l's in the 15mbps and u/l's in the 130mbps range. Several test sites, both OS X and Windoze produce the same results.

iperf, using UDP, produced results in the specs. d/l and u/l in the 290mbps range. I'm wondering if there maybe something related to residential vs commercial. I'm thinking that I may just need to build an iperf setup for testing and ignore the websites.
 
Not much help here, but I have a client running redundant Sophos XG 210's, managed by a 3rd party company on a gigabit connection and most speed test sites won't even load. Sophos is too strict on some things. We can only get 300mbps through the firewall, according to Sophos it has to do with the speed test only using 1 core for the connection.
 
I haven't had problems with 200Mbps symmetrical on speedtest.net, so the threshold to where you start getting problems must be higher than that.
 
Thanks for the responses. Guess it's just the nature of the beast. I do know that there is some variability in results but circuits over 100mbps have been way off. I guess I'll just have to dump the rest and just use iperf as a benchmark.
 
TBH, and I know it sucks..
A good newsgroup, Fast File sharing (Torrents, etc) are sure fire ways to determine your speed. Since it is a distributed, multi-connection download - if you have enough peers, you can saturate your connection. I usually fire up a torrent download of Ubuntu since there are a lot of shares for that. If your download is too fast for a simple 4.5GB then add a torrent for each of the ISO's so that your torrent client is loaded up.
 
Haven't had problems with todays faster connections.

If the ISP has a speed test, I often use that first. It tests between the site you're at, and something inside that ISPs data center. This will determine the connection between your location and the ISPs data center. NOT the internet yet, you're not going out the ISPs main gateway out into the world yet.

And then step outside to public servers to test no only your ISPs connection to their gateway, but...beyond that and out on the public internet.
I'll run several different tests, speedtest.net being a typical first one.
testmy.net is a decent one for watching a longer running steady test...get that nice bar graph which illustrates more of a steadiness of performance.

I'm intrigued by that speedof.me that was mentioned in these forums not long ago...I like how it throws variables in there like file sizes and types. And its results should be treated as such.

FYI...Google has one just for gigabit fiber...however, it always seems to be based out of the middle of the US, I have never seen an east coast server come up for me. But may be another one to throw in your regular mix of tests.
http://speedtest.googlefiber.net/
 
You must log in or register to reply here.

Similar threads

tankman1989
Earn extra revenue by switching business clients to quality ISP's - referral fees!
Replies
7
Views
1K
tankman1989
tankman1989
SilverLeaf
Hardware Testing with Parted Magic
2
Replies
22
Views
31K
SilverLeaf
SilverLeaf
Back
Top