[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Repository services and budget
- Date: Wed, 8 Jan 1997 19:33:13 -0800 (PST)
- From: Kent Crispin <kent@songbird.com>
- Subject: Re: Repository services and budget
Jeff Williams allegedly said:
>
[...]
> > I wouldn't. I would keep a duplicate spare. Two pentium machines.
> > Maybe $10000.
>
> Ok, that should be fine. BUt they will need to be sharing drives at
> least.
For the ultimate in reliability, perhaps. But with the scheme I'm
using I don't think it is necessary. But it certainly wouldn't hurt.
> Well I was not talking about the machine being down, rather I was
> refering to a bad drive or a scheduled archive. There should be
> no reason that the DNS is compleatly down.
The central server (which I called the "master") doesn't archive
anything. The average lifetime of data items on its disk should be
less than a couple of hours. If it goes down for more than two hours
everything will have expired, so there is no point in rebuilding the
data. So what you do is keep it down for two hours when it goes
down, so no conflicting certificates are generated.
> Where are the Domain def's kept than? Where is the DNS db itself?
> On a drive somewhere, correct? I agree that there should be multipul
> servers for the zone. But you did no include that in you previous post.
Nobody has so far talked about the actual nameservers. In all
scenarios the DNS servers have to be reliable; their databases files
have to be carefully maintained, etc. The algorithm I use
capitalizes on the fact that we already must have a reliable, fault
tolerant DB, namely DNS, available to us. Once things are in DNS it
is a requirement of *any* registry model that they be reliably
maintained.
--
Kent Crispin "No reason to get excited",
kent@songbird.com,kc@llnl.gov the thief he kindly spoke...
PGP fingerprint: 5A 16 DA 04 31 33 40 1E 87 DA 29 02 97 A3 46 2F