Intermittency during heavy upload?

I’m at my wits end here so I'm reaching out to see if anyone else has experienced this or anything like it. After the 15.1.1 update we have been getting trouble calls daily about intermittency. We now run almost exclusively 450M AP, with some 450I, most of it in the 5.8. the 15.1.1 update rendered the 5.4 almost completely unusable in our area. At first we thought it was just customers maxing out their bandwidth, but as we looked closer and continue to get trouble calls daily about this, we believe there is a problem. We just cant figure out what it is. This is what we have been able to recreate. Example, a customer has a speed package of 6M down 2M up sustained, Burstable to 10M for 10M down, and 3M for 3M on the up. The link test is good, speed test are good, everything looks great, but when the customer maxes out the sustained upload, traffic appears to stop or become so slow that everything else trying to access the internet just times out. That one upload still runs fine, but nothing else will run at all. We have recreated this on our low package 1.5M-384k, as well as our high one at 15M – 5M. The problem became most evident when we moved over our last 100, PMP 100 customers to the 450M and we began receiving daily calls asking to go back to the old system because it worked so much better. At this point we are loosing customers every week due to this. All the config in the AP and SM is the same as before the update. The problem started with the 15.1.1 ver. Am I missing a new setting? Some kind of configuration issue? Or should I just revert back to 15.0.1, the last stable ver that we had no problems with? Would updating to 15.1.2 Help? I don’t see anything in the notes about a problem like this. We just spent $500,000 on moving our network over to the 450 series, and now our service is worse than ever, and we are loosing customers weekly. Any help or ideas are greatly appreciated.

For every version I can remember, if the upload MIR is saturated the download will suffer greatly. This is usually caused by google drive, or drop box consuming all of the upload bw. This doesn't seem to be as much of a problem, or at least not noticible on the ePMP platform, maybe MIR rates are not nearly as strict as on the PMP450 platform?

When a client calls in reporting download issues, we run a deep protocol and bandwidth report using SandVine and can very quickly tell what the client is doing... often times, they don't even know that someone in the house is completely tapping their connection with one of these services. It sounds like you have a pretty big client base and network... have you considered investing in a box to handle traffic flow and reporting?

Lastly, there are a number of QoS improvements coming out to the upcoming 15.2 release that might help with this situation.