26.07.2013 Views

WiMax Operator's Manual

WiMax Operator's Manual

WiMax Operator's Manual

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

CHAPTER 7 ■ SERVICE DEPLOYMENTS OVER PUBLIC WIRELESS MANS 167<br />

protocol because it merely signals the requirements for various traffic flows. Incidentally, RSVP<br />

requests must reach every router or switch in every conceivable signal path the packet may<br />

take, so it tends not to scale well over distance. RSVP does not employ extra headers or labels,<br />

but it frames its requests for bandwidth within the normal IP header in the form of priority bits.<br />

IntServ, DiffServ, and IP Precedence<br />

IntServ stands for integrated services and was a protocol developed by the IETF in the 1990s. It<br />

is little used today, and I will pass over it here.<br />

DiffServ, which is widely used today, performs a somewhat different function than RSVP;<br />

it provides a classification scheme for various streams rather than requesting bandwidth along<br />

a data path. DiffServe, as well as RSVP, requires that priority bits be assigned to the packet<br />

header to establish the priority of that packet in terms of the queue it will occupy and the order<br />

in which it will be transmitted relative to packets of other priorities occupying other queues.<br />

Such queuing will take place at switches rather than routers, whether MPLS switches (increasingly<br />

the case today), ATM switches, or metro Ethernet switches.<br />

As is the case with RSVP, all relevant network elements must be enabled for DiffServ, or<br />

QoS will not be maintained across the network.<br />

IP Precedence fulfills a similar function to DiffServ, but the latter is much more widely<br />

deployed today. IP Precedence, unfortunately, is not consistently implemented from one<br />

vendor to another and thus is of limited usefulness in real-world networks.<br />

RTP and RTSP<br />

Real Time Protocol (RTP) and Real Time Streaming Protocol (RTSP) are more specialized than<br />

the other protocols covered thus far, being designed primarily to support audio and video<br />

streaming. RTP, which has somewhat broader application than RTSP, is used for a number of<br />

real-time applications, including but not limited to audio and video. RTSP is used mainly in<br />

multicast or media-on-demand applications and in IP telephony as well. RTP is a transport<br />

protocol, and RTSP, as the name implies, is as streaming protocol used for transient singlesession<br />

multimedia presentations rather than the transfer of storable multimedia files. RTSP is<br />

designed specifically to work in a client-server environment.<br />

MPLS<br />

MPLS is a switching protocol that supports QoS by permitting the setup of label-switched<br />

paths for traffic of similar characteristics. It is not, as I have seen, intended primarily for QoS<br />

but rather for effective network management.<br />

Queuing Techniques That Support QoS<br />

In addition to the previous protocols, which primarily involve signaling, a number of protocols<br />

promote QoS by managing the relationship between and among queues assigned to different<br />

classes of traffic. These include strict priority queuing, round-robin queuing, fair queuing,<br />

weighted fair queuing, adaptive scheduling, and so on. The first requires the complete emptying<br />

of the highest-priority queue before the switch proceeds to that which is next in priority,<br />

and the others strive to provide some degree of equity to lower-priority streams. No such protocol<br />

can be said to be perfect, and network operators will have to determine which will best<br />

satisfy their customers in aggregate. Normally these queuing protocols will not be the province

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!