[2438] in SIPB_Linux_Development
Re: remaining issues for 8.2
daemon@ATHENA.MIT.EDU (Mike Whitson)
Fri Jan 29 14:07:03 1999
To: Emil Sit <sit@MIT.EDU>
Cc: linux-dev@MIT.EDU
From: Mike Whitson <mwhitson@MIT.EDU>
Date: 29 Jan 1999 14:06:46 -0500
In-Reply-To: Emil Sit's message of Fri, 29 Jan 1999 07:04:01 EST
Emil Sit <sit@MIT.EDU> writes:
> Would there be any load issue with having a lot of machines directly
> synchronizing with tick?
In general, for scalability reasons, it's not a wonderful idea to have
lots of clients pointed directly at stratum 1 servers such as tick.
We'd prefer that if you need to have a general client configuration,
that clients synchronize with the Athena-cell DB servers: agamemnon,
chimera, orf, and prill. These are kept at stratum 2; each is served
by tick and a different off-campus stratum 1, and they all peer with
each other. This makes the pool of them fairly reliable and robust at
keeping stratum 2.
-wrong mike
(with the ASO Timekeeper hat on)
