Link test not accurate under load

Anyone else notice this?  I had a tech call me because the last several installs his link test was not hitting 10x5 which is the qos plan we put most of them on. The one today had a -58 and 14 mcs rates and looked real good.  His link test was all over the place, but mostly 5x5.  I noticed on the throughput that he was using anywhere from 3megs to 10 constantly from his laptop.  (Who knows what hes got installed on that thing).  I then had him unplug his laptop and the link tests were consistly 10x5 at that point.

Im using 3.5.1 on aps and sms.

the link test is accurate, it will not dump customer traffic to provide a result.  it will show you the reaming bandwidth on a busy link. 

when the link is unloaded, you'll see the score you're looking for. 

3 Likes

Good to know its suppose to work like that.  Coming from the old canopy days I prefer the old method so you could really guage how healthy a link is.  A 4 second data drop when a customer is on the phone with tech support has never been a problem.  

I guess it works the same way with PTP links as well?  We havent started using ePMP for PTP yet, but that would be a little on the irritating side.  Maybe a future release can give us the option of either or?  Either that, maybe the module could do a calculation for us and show the total of used vs available?

On the 450 side, we are actually implementing this type of link test, internally dubbed a "flood test" because it just floods the remainder of the pipe, but gives you the total throughput capable as a result.

The key there is that it is not "service interrupting", which has been the major complaint about the traditional Canopy Link Test.

1 Like

You can do a 20 second test then look at your throughput graph, not on the same page but you can see the metric

2 Likes