Content deleted Content added
No edit summary |
No edit summary |
||
Line 20:
No, the leaky bucket and the token bucket are symmetrical, and will see the same packets as conforming or nonconforming, if given the same parameters. The only real differences are that the leaky bucket description is about adding water when packet conformance is tested and taking it out at regular intervals; whereas the token bucket is about taking out tokens when conformance testing and adding them in regularly. Functionally they are identical. Yes I have read Tanenbaum's description in his book computer networks. However, I've also read Turner, who Tanenbaum credits as the first to describe it, and the ITU-T/ATM Forum description in I.371 and the Forum's UNI. Sorry for this, but Tanenbaum is wrong; or at least his description is simplified to the point where it is, in general, wrong, and applies only to a single specific case, where the bucket depth is exactly equal to the water added by a packet or cell. Only in this special case does the leaky bucket not alow for any jitter or burstiness, as does the token bucket when its depth is equal to the number of tokens required for a packet. Sorry, I'm getting stressed by the repatition of this particular falacy, can you tell?
Graham Fountain 14:52, 9 May 2010 (UTC)
I have added a new section in the discussion page of the leaky bucket algorithm that adds to this issue.
Graham Fountain 23:20, 5 June 2010 (UTC)
| |||