[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[RRG] Questions: Push/pull, ECMP



Hello IRTF folks,
Somewhere it started: Push or pull model. Likewise, an older issue: ECMP.
 
Though I have some idea what might it be (push/pull) or how it may be done (ECMP algorithm),
I would appreciate very much if some one of the IRTF folks provided the correct definition/solution
for these issues.
 
As you see, I picked some RAM-email where push/pull is adressed. Maybe, if  we all get a clear understanding then we may have an appropriate technical discussion. Religion is for more important issues, imo.
 
Thanks in advance
Heiner
 
 
In einer eMail vom 09.05.2007 18:40:40 Westeuropäische Normalzeit schreibt drc@virtualized.org:
Eliot,

On May 9, 2007, at 9:13 AM, Eliot Lear wrote:
> I feel like we've had this argument before.

Yep.  You and others keep asserting that a pull-based mapping 
distribution model is a "non-starter" because it "won't work".  I've 
been trying to get from folks _concrete_ examples of applications 
that "won't work" in the face of increased latencies on the order of 
10s or 100s of milliseconds after a cache miss.  In response I get 
vague handwaving about voice or video (I know from personal 
experience that UDP-based NFS works fine in the face of an initial 
packet delay of 10s of milliseconds) or descriptions about classes of 
applications that would appear to be degenerate cases and asymmetric 
routing that would likely have unpredictable performance 
characteristics with pull or push.

GIven the lack of data, it would seem we've descended to the realm of 
religion, so I'll give it a rest.

Rgds,
-drc


_______________________________________________
RAM mailing list
RAM@iab.org
https://www1.ietf.org/mailman/listinfo/ram
 
--- Begin Message ---
Eliot,

On May 9, 2007, at 9:13 AM, Eliot Lear wrote:
I feel like we've had this argument before.

Yep. You and others keep asserting that a pull-based mapping distribution model is a "non-starter" because it "won't work". I've been trying to get from folks _concrete_ examples of applications that "won't work" in the face of increased latencies on the order of 10s or 100s of milliseconds after a cache miss. In response I get vague handwaving about voice or video (I know from personal experience that UDP-based NFS works fine in the face of an initial packet delay of 10s of milliseconds) or descriptions about classes of applications that would appear to be degenerate cases and asymmetric routing that would likely have unpredictable performance characteristics with pull or push.

GIven the lack of data, it would seem we've descended to the realm of religion, so I'll give it a rest.

Rgds,
-drc


_______________________________________________
RAM mailing list
RAM@iab.org
https://www1.ietf.org/mailman/listinfo/ram

--- End Message ---