Re: [Fwd: RE: LC: draft-ietf-forces-protocol-10.txt>>ForCES message size?]

I agree that it's not necessary to segment ForCES message,but we should 
have a way to avoid a sender to send an arbitrary size messages which will 
be problematic for a receiver.Shined by SNMPv3,I suppose we could add a 
field in ForCEScommon header named maxMessageSize which is set by the 
sender of the message and means the max message size the sender can 
accomodate,the value can be determined by the specified implementation. 
Yours,Fenggen
>From: "Joel M. Halpern" <[email protected]>
>Reply-To: "Joel M. Halpern" <[email protected]>
>To: [email protected]
>Subject: Re: [Fwd: RE: LC: draft-ietf-forces-protocol-10.txt>>ForCES 
message size?]
>Date: Tue, 12 Jun 2007 08:14:45 -0400
>
>I would strongly prefer not to complicate the protocol by defining a 
>size limit and then trying to figure out what to do if the messages 
>are too big.
>The Redirect channel is supposed to be congestion controlled.
>Also, since redirect packets ought not be reassembeled (they ought 
>to be the packet as received on the interface by the fE) it follows 
>that they are in practice size limittd well below the IP theoretical 
>limit of 64K.
>Note that for response messages, if the requestor cares to limit 
>sizes he can get things in pieces that will get that effect.  (i.e. 
>the CE can choose how much information to ask for.  If it wants a 
>whole table, but wants to control possible size issues, it can 
>request the data in pieces.  Or even one entry at a time.)
>For notifications, it seems that good notification design will not 
>put too much information in any one notification.  But I would 
>strongly prefer not to have a defined maximum that everyone must 
>stay under.
>
>Yours,
>Joel
>
>At 04:30 AM 6/12/2007, Patrick Droz wrote:
>>I read the protocol draft again and still want to raise my early
>>question:the size of different ForCES message,i think not restrain 
>>it and
>>let implementation to freely choose will result in disasters,the 
>>reason is
>>that the protocol implementator not sure what message is legal and 
>>must
>>handle all sizes,a attacker may send very large size message to 
>>cause the
>>receiver down.I suppose we should limit the size of each ForCES 
>>message
>>accordingly.(FYI:The BGP set the message size limit of 4096 bytes, 
>>our
>>ForCES message is different,we have two sort of message,redirect 
>>message is
>>one sort,and all other messages belongs to the other,we should 
>>treat them
>>diffierently.)
>>Yours,Fenggen

_________________________________________________________________
璐逛杞 MSN Explorer:   http://explorer.msn.com/lccn/