ONT - Integrated Service Model

bighornsheep

What would be an example of QoS that is in the Integrated model?

In Cisco Press ONT Exam Cert. Guide by Ranjbar:

IntServ mimics the PSTN model where every call entails end-to-end signaling and securing resources along the path...Controlled volume services within the Cisco IOS QoS feature set are provided by RSVP and advanced mechanisms such as LLQ. The Guaranteed Rate service type is offered by deploying RSVP and LLQ.

in a typical DiffServ network, voice traffic is assigned to a priority queue that has reserved bandwidth (through LLQ) on each node.

So does that mean LLQ is both Integrated & DiffServ? What would be another queueing/QoS feature under Integrated model?

Thanks for any help![/quote]

kryolla

Int Serv model is used for VOIP so when you want to place a call all the routers will signal each other to reserve bandwidth for the call and once all routers are able to make the reservation the call is placed. The command under interface mode is ip rsvp bandwidth

networker050184

kryolla wrote:

Int Serv model is used for VOIP so when you want to place a call all the routers will signal each other to reserve bandwidth for the call and once all routers are able to make the reservation the call is placed. The command under interface mode is ip rsvp bandwidth

The Int serve model is not limited to VoIP and not all VoIP implementations (none that I have seen actually) use the Int serve model. All the Int serve model does is guarantee end to end reservation of resources while Diff serve is on a device by device basis but lays out a frame work for all devices to use.



To the OP I don't believe LLQ would be classified as Int serve because it does not guarantee bandwidth end to end, just on the interface where it is applied. That is the premise of the Diff serve model. I could be wrong though. Will let you know soon when I take the ONT

APA

LLQ is part of the DiffServ Model but can also be deployed with RSVP as part of the IntServ Model.

DiffServ model does not guarantee absolute bandwidth reservation, LLQ however does provide a specific traffic class guaranteed bandwidth over an interface during periods of congestion... It is a strict-priority queue which is policed during congestion to avoid starvation of other flows....

IntServ uses RSVP for signalling to guarantee BW along the whole path for requesting apps...if the whole path cannot guarantee this bandwidth then the application cannot begin transmission....

You can deploy RSVP & LLQ together to allow guaranteed-rate service as mentioned by Ranjbar in the Exam Cert Guide..... This allows the signalling and BW reservation along the whole path via RSVP but also the benefits of low delay and pre-defined BW reservations for specific traffic classes via LLQ.

What if an app requests BW during a congested period and gets turned down by RSVP due to limitation along the path???? With LLQ enabled you could assure that whenever this app requests BW via RSVP along the path it is always allocated BW aslong as it doesn't exceed it's reserved amount.

Also IntServ is not limited to VoIP traffic flows.... Neither is LLQ..... If you have a business critical application that is just as important as Voice traffic by all means give it priority over other traffic....

It's just that in most converged networks today Voice is the most time-sensitive traffic flow you will come across.... (Possibly Video as well nowadays though...)

kryolla

Nobody uses the int serv model because it is legacy and does not scale well. The op wanted an example and the most realistic is VOIP. LLQ is not limited to voip or video but the norm is to put delay sensitive traffic in a priority class. QOS is not difficult the most complicated part of QOS is traffic shaping and policing.

bighornsheep

Thanks for all of your input, I think I get the IntServ model using LLQ+RSVP now....so signaling for resource reservation end-to-end is the concept of IntServ, but using LLQ for the signaling traffic is to guarantee bandwidth for the signaling flow.

kryolla wrote:

QOS is not difficult the most complicated part of QOS is traffic shaping and policing.

Would you care to explain the difference? I know that the priority queue within LLQ is said to be "policed" because it doesn't starve other queues, what's the mechanism behind this policing? What would be considered traffic shaping?

Is it me or does Ranjbar's ONT book seem to be missing quite abit on the ONT exam objectives? There seems to be so much I don't know even after the book.....any other recommended resources?


Thanks for all of your help.

redwarrior

For ONT, I found the CBT Nuggets + Trainsignal videos to really help filling in the blanks. The exam guide along just didn't cut it.

As far as understanding DiffServ and Intserv and how LLQ fits in, the way I thought of it was this, and it could very well be wrong. With Intserv, you use RSVP to reserve bandwidth across every hop in the packet's path and that's it. With DiffServ, you're concerned with PHP (per Hop Behavior) and you use different QoS mechanisms to guarantee a Service Level Agreement. LLQ is one of those mechanisms available and you implement it on a single hop to ensure the behavior on that hop. IntServ delivers end-to-end QoS by simply reserving bandwidth across the entire path without taking into consideration different conditions at each hop. DiffServ delivers end-to-end QoS by marking packets and then letting you determine on a hop-by-hop basis how those packets will be treated. That means more configuration for you, but also a lot more control.

Of course...I could be completely smoking the smack here.