Thursday, July 31, 2008

QoS exam 642-642

I'm feeling pretty good about this exam and will take it either next week or the following week. It's amazing how much LESS material there is in a professional level exam vs an expert level one.

As there's a lot of overlap with my R&S CCIE and my previous voice experience, I'm not stressing out much over this. Of course, there's always the fear of taking it for granted and failing and wondering how the heck I can pass the CCIE written and Lab and bomb the QoS exam.

Anyway--here are the topics I need to commit to memory:
  • Higher drop precedence means more likely to be dropped. But higher IP precedence means less likely do be dropped. Huh? Who designs this stuff anyway? Just remember, AF11 will be dropped before AF21; however AF12 will be dropped before AF11. That last number means drop precedence, as in higher priority to be dropped. Whereas the first number means the traffic itself is higher priority, and less likely to be dropped. It makes me think they designed it this way simply because they knew they'd get a few good test questions out of it.
  • RTP Header Compression also compresses the UDP and IP headers, and therefore must be enabled on both sides of a link. Could the name be any more misleading?
  • COPS, aside from one of my wife's favorite TV shows, is a protocol that allows centralized monitoring and control of RSVP. Bad packets, bad packets, what'cha gonna do? What'cha gonna do when QoS comes for you?
  • ECN is like frame-relay adaptive shaping, except it's not shaping and there's no frame-relay. If a router is nearing congestion, it can use the last two bits of the ToS field to notify the upstream host. Then the upstream host relays this information to the sender. Finally, the sender slows down sending. It's like FECN and BECN, and like being issued a warning before WRED steps in and destroys your packet.
  • BGP can use communities to propagate QoS policy information (bgp-policy command)

No comments: