Sowmini.Varadhan at Sun.COM writes: > On (02/27/09 09:46), James Carlson wrote: > > And I'm not buying any "performance" argument here. We're talking > > about the delivery of errors. Applications aren't built to deliver > > bulk data via error reports. If performance (or perhaps more > > accurately DoS risk) is really a concern, then we should just throttle > > the rate at which we send any messages at all, rather than trying to > > We already have icmp error rate-limiting in place for that.
Indeed. My point is just that optimizing the byte count here is no win for performance. > > So, yes, I think the default should be larger, but I'm really unsure > > that there should be a tunable here. > > exactly. New tunables like these should come with a "best used by" date. For those introduced in 1999 and never used, it's probably well past its prime. Some quick internal Sunsolve searches should help answer any questions about whether customers were told to use them. -- James Carlson, Solaris Networking <james.d.carlson at sun.com> Sun Microsystems / 35 Network Drive 71.232W Vox +1 781 442 2084 MS UBUR02-212 / Burlington MA 01803-2757 42.496N Fax +1 781 442 1677
