Re: (virtual) hum on CARD open issues
"Eunsoo Shim" <[email protected]>
| Newsgroups | gmane.ietf.seamoby |
|---|---|
| Message-ID | <014601c34fc6$e4e18650$c96b0f8a@peace> |
Vijay, Please see my comments below. > > I already provided the text when I reviewed the CARD protocol. > > The MN MUST send only one CARD Request per > CARD_RETRANSMISSION_INTERVAL and not more than CARD_MAX_RETRIES. > If the MN sends requests more frequently, the AR SHOULD drop the > CARD requests and not process them. > > and add to the constants section > > CARD_RETRANSMISSION_INTERVAL 1 second > CARD_MAX_RETRIES 3 > > this is similar to ICMP rate limiting. > > TCP rate limiting is different, not applicable here. in general > in any protocol where there is a request and a reply (and no > more messages), ICMP rate limiting is best suited. you dont have > to "source-quench" the Mobile Node. :) > What's going to happen if the AR should deal with a large number of MNs? Also what if the network operator wants to use a different value for the inter-message interval limit? How does the MN know the value? You might think "source-quench" is significant additional complexity but actually the AR must calculate the message rate of each MN to do policing (that is, dropping messages over the rate limit) and thus doing "source-quench" adds little additional complexity. Eunsoo