Community discussions

MUM Europe 2020
 
syadnom
Member
Member
Topic Author
Posts: 421
Joined: Thu Jan 27, 2011 7:29 am

stacked queues, some advice/help

Wed Aug 27, 2014 3:16 am

I'm looking to change things up a bit. Currently using simple queues and buying enough bandwidth to handle it, but that means smaller DL numbers for clients. I want to move beyond the typical 3-5Mbps plans.

I'd like to do a stacked queue, so let me describe it and someone can tell me if this is possible with 'tik. I'm happy to get a more powerful x86 or CCR router to.

Basically, I want to create a top level queue (MASTER) that is 90% of my uplink (currently 150Mbps but I've gigabit to the provider and can upgrade in about 15 minutes).

I want to move from a hard 3x1 type setup, to a tiered bursting system. So I'll offer (for example) a 6x1 and 18x3 with a 3x512k commit on all consumer plans.

Now I'd like to create a dedicated queue for VoIP (queueVOIP) with high priority with previous queue set to master.
Then another queue for other traffic that is a PCQ (queueDATA), so every single queue under it will evenly split all available data.

below queueDATA I would put individual queues of individual customers.

something like
MASTER
|-queueVOIP (high priority)
|-queueDATA (regular priority)
.|-user1queue
.|-user2queue
.|-user3queue


I know the upstream addresses of all voip traffic I'll identify, it's my own pbx in the datacenter or my providers sip switch, so I can tag voip and related connections so all customers VoIP hardware will be outside their regular plan speeds and will be higher priority.

I don't care about other traffic types right now, I just want everyone to fall into a fair sharing setup.

In other words, I buy 150Mbps and I want to be able to give the excess to customers. Also, I'm going for near-neutral net and only shaping to build plans and ensure quality voice calls.

I'll additionally be bulk shaping at each pop, basically just setting some ToS and prioritizing voice, limiting traffic to the link's capabilities for low-latency.

Thoughts/.

Who is online

Users browsing this forum: No registered users and 108 guests