From: Howard C. Berkowitz (hcb@xxxxxxxxxxxx)
Date: Wed Mar 13 2002 - 21:30:21 GMT-3
At 11:23 PM +0000 3/13/02, Brian Lodwick wrote:
>Interconnections
>by Radia Perlman
>
>>>>Brian
>
>
>>From: "A.Steinbock" <ccie10@usa.net>
>>Reply-To: "A.Steinbock" <ccie10@usa.net>
>>To: <ccielab@groupstudy.com>
>>Subject: Reliable UDP?
>>Date: Sun, 10 Mar 2002 22:52:44 -0800
>>
>>Back to Basics:
>>Isn't it true that TCP is connection-oriented & reliable and UDP is
>>connectionless & best-effort?
>>
>>Reading about Cisco IDS I came across the following paragraph:
>>
>>"Communication between the Cisco Secure IDS Sensors and management
>>consoles is
>>provided by a proprietary, reliable UDP transport protocol that guarantees
>>transmission to intended recipients and maintains connection status with all
>>of the IDS components."
>>
>>http://www.cisco.com/warp/public/cc/pd/sqsw/sqidsz/prodlit/ids_qa.htm
>>
>>UDP, Reliable, and guarantee?
>>This sounds more like TCP than UDP.
>>Could some one help me out of this confusion please?
This was probably written by the same marketdroid that, when Sprint
implemented GSRs, put out the press release boasting of increased
service and increased latency.
Now, in fairness, there are such things as acknowledged datagram
protocols -- LLC type 3, Appletalk Transaction Protocol, X.25 Fast
Select, among others, but I doubt this is what the writer had in
mind. These are not strictly connection-oriented or connectionless.
A better term would be that they are stateful but connectionless.
Things get even more complex when you start differentiating between
hard state (an explicit disconnect), soft state (IGMP and RSVP, for
example), and soft state with optional explicit teardown.
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:57:04 GMT-3