Re: [RRG] Why delaying initial packets matters

["William Herrin" <bill@herrin.us>]

On Thu, Feb 21, 2008 at 9:43 PM, Brian Dickson <briand@ca.afilias.info> wrote:
>  The short list of things that I think need further testing or analysis
>  or experience, regarding loss
>  or delay to the first packet:
>  ntp (think about it!)

Brian,

Say, that's a good one.

NTP clients querying servers through a TRRP path with oscillate the
servers between eligible and ineligible each time the delay between
NTP queries gets long enough to cause the address to time out of the
ITR causing what the NTP client will perceive as excessive delay and
jitter.

Solving this in the short term requires no changes to NTP but it does
require some attention to configuration for those who choose to be
early adopters of TRRP space. Far-away NTP servers won't be usable
over a TRRP path so you'll need a machine on BGP routable space to
make those far-away queries and then you'll have to query that machine
over the local routing infrastructure that doesn't pass through BGP or
TRRP. Fortunately, TRRP's ETRs already meet this criteria and many
machines capable of running a GRE endpoint are also capable of running
NTP.

In the long term, NTP client software could be adjusted to notice the
unexpected delay phenomenon and either shorten the time between
queries to the point where the destination stays in the ITR's cache or
recognize the delay as transient and retry a few seconds later
ignoring the transient delayed result. The ITR will generally only
have to re-cache the bottom-level component, so that cache oscillation
shouldn't be harmful.

This will be an issue for every map-encap scheme that relies on caching.


>  snmp
>  snmptrap

Generally local; outside TRRP's area of effect.

>  l2tp

Don't know for sure but I don't think there's a problem with l2tp or
pptp over TRRP-managed address space. VPN protocols are generally
already designed to work over nasty, unpredictable networks.

>  all the various *lm things (licence managers) - how many IT help desks
>  to we want to anger at once, really?

Generally local; outside TRRP's area of effect.

Regards,
Bill Herrin


-- 
William D. Herrin herrin@dirtside.com bill@herrin.us
3005 Crane Dr. Web: <http://bill.herrin.us/>
Falls Church, VA 22042-3004

--
to unsubscribe send a message to rrg-request@psg.com with the
word 'unsubscribe' in a single line as the message text body.
archive: <http://psg.com/lists/rrg/> & ftp://psg.com/pub/lists/rrg