|
Home > Archive > DNS Bind > February 2004 > question about caching & bind
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
question about caching & bind
|
|
|
| Hi,
I currently run a split DNS scenario and was wondering about caching
to possibly improve performance. Is it possible or even advisable to
have 1 or more DNS servers (say internally for arguments sake) utilize
another separate bind caching only server ? My thinking is that if one
or both of the servers is rebooted I lose the rich cache it has
created and it would be nice to prime the lost cache quicker. If so
what is considered best practice ?
thanks,
Joel
| |
| phn@icke-reklam.ipsec.nu 2004-02-17, 5:33 pm |
| joe <bonggo@nospam.rocketmail.com> wrote:
> Hi,
> I currently run a split DNS scenario and was wondering about caching
> to possibly improve performance. Is it possible or even advisable to
> have 1 or more DNS servers (say internally for arguments sake) utilize
> another separate bind caching only server ? My thinking is that if one
> or both of the servers is rebooted I lose the rich cache it has
> created and it would be nice to prime the lost cache quicker. If so
> what is considered best practice ?
Keeping nameserver(s) running and avoiding forwarding might be
considered "best practice"
"rebooting nameserver" ?? If you are thinking of toy-operating systems,
there is other places to discuss "how often" and "why" rebooting before crash.
> thanks,
> Joel
--
Peter Håkanson
IPSec Sverige ( At Gothenburg Riverside )
Sorry about my e-mail address, but i'm trying to keep spam out,
remove "icke-reklam" if you feel for mailing me. Thanx.
| |
|
| I appreciate your reply, although the "toy-operating systems" comment
was really unnecessary. It happens, servers get re-booted, its a fact
of life, but I still manage to sleep at night. I realize keeping the
server up is the main goal, however my question still stands. Can I
re-prime the cache as explained or is this is wasted effort.
regards,
Joel
On Wed, 18 Feb 2004 06:57:00 +0000 (UTC), phn@icke-reklam.ipsec.nu
wrote:
[color=blue]
>joe <bonggo@nospam.rocketmail.com> wrote:
>
>
>Keeping nameserver(s) running and avoiding forwarding might be
>considered "best practice"
>
>"rebooting nameserver" ?? If you are thinking of toy-operating systems,
>there is other places to discuss "how often" and "why" rebooting before crash.
>
| |
| phn@icke-reklam.ipsec.nu 2004-02-18, 7:33 am |
| joe <bonggo@nospam.rocketmail.com> wrote:
> I appreciate your reply, although the "toy-operating systems" comment
> was really unnecessary. It happens, servers get re-booted, its a fact
> of life, but I still manage to sleep at night. I realize keeping the
> server up is the main goal, however my question still stands. Can I
> re-prime the cache as explained or is this is wasted effort.
> regards,
> Joel
The cache will reprime itself, and remember that most data in the cache
is less then a day old ( does anyone have figures of most common TTL
values ?) .
Rebooting has no value in itself, you have everything and your uncle
to gain from keeping your bind 'ticking'
Like this one :
bind 88174 0.0 15.3 80552 79176 ?? Ss 25Oct03 68:01.72 named -t /var/named -c /named.conf -u bind -n 2
^^^^^^
running bind-9.2.3 , downloaded and built that same day ...
the machine self has been up :
> w
9:43PM up 181 days, 13:58, 1 user, load averages: 0.00, 0.00, 0.00
Of cource it's running a "brand OS" designed for reliability.
--
Peter Håkanson
IPSec Sverige ( At Gothenburg Riverside )
Sorry about my e-mail address, but i'm trying to keep spam out,
remove "icke-reklam" if you feel for mailing me. Thanx.
| |
| Barry Margolin 2004-02-18, 8:33 am |
| In article <c108jf$273c$1@sf1.isc.org>,
joe <bonggo@nospam.rocketmail.com> wrote:
> I appreciate your reply, although the "toy-operating systems" comment
> was really unnecessary. It happens, servers get re-booted, its a fact
> of life, but I still manage to sleep at night. I realize keeping the
> server up is the main goal, however my question still stands. Can I
> re-prime the cache as explained or is this is wasted effort.
Unless the reboots happen frequently, it's probably not worthwhile
designing your server architecture to deal with this. Will anyone
really notice an extra second of delay that only happens once every few
months?
The fact that you asked the question suggested that frequent reboots
*is* a problem for you, implying that you're using an unreliable (i.e.
"toy") OS on important servers. You'll get much more bang for your buck
by improving general reliability than by trying to design workarounds
for the problem.
--
Barry Margolin, barmar@alum.mit.edu
Arlington, MA
| |
| David Botham 2004-02-18, 9:33 am |
| bind-users-bounce@isc.org wrote on 02/18/2004 03:45:32 PM:[color=blue]
> joe <bonggo@nospam.rocketmail.com> wrote:
No, you can't, not practically.
Dave...
| |
|
|
you might be able to "re-prime" the cache from a non-auth
source, but itis much better practice to get the data from
authortative sites. e.g. nearly everyone will tell you
this is wasted effort.
--bill
>
> I appreciate your reply, although the "toy-operating systems" comment
> was really unnecessary. It happens, servers get re-booted, its a fact
> of life, but I still manage to sleep at night. I realize keeping the
> server up is the main goal, however my question still stands. Can I
> re-prime the cache as explained or is this is wasted effort.
>
> regards,
> Joel
>
> On Wed, 18 Feb 2004 06:57:00 +0000 (UTC), phn@icke-reklam.ipsec.nu
> wrote:
>
>
>
| |
| Pete Ehlke 2004-02-18, 2:33 pm |
| > >[color=blue]
The "benefits" of large caches are mostly a myth. There is very
interesting work (see, for example
http://nms.lcs.mit.edu/papers/dns-ton2002.pdf esp. Section V,
"Effectiveness of caching") showing that in the real world, very, very
small numbers of names account for the vast majority of queries, and
that of the remainder, a significant percentage are referenced only
once. In other words, your cache quickly reaches most of its potential
richness very quickly, even with very few clients, and many of the
non-cache lookups would not be in cache no matter *how* large and
robust.
Which is a long winded way of saying "yes, it's a wasted effort".
-Pete
|
|
|
|
|