BIND 8.2.3 verus 9.x.x ?? in production

[Brad Knowles]

At 11:31 AM -0700 3/20/01, Nate Duehr wrote:

>  I agree with Kevin, but it has more to do with the 41,000 queries one of
>  my servers took in 40 seconds last weekend than anything.  :-)

	Ahh, but that's only a thousand queries a second.  I had machines 
back in 1996 that could do that many.  ;-)

>  I also think calling it BIND 9 is a misnomer.  It's NIND 1.1.1rc5,
>  not BIND 9.1.1rc5.  (NIND = Nominum Internet Name Daemon -- Berkeley's
>  not really all that involved anymore, are they?)

	No, this is not at all correct.  BIND is an ISC product, and 
Nominum's involvement is as the company that has been contracted to 
write the code.  Therefore, this is very much still BINDv9, since 
this is still an ISC project.

>  When you look at it in that light... I'll stick with 8.x.x for now.

	Keep a close eye on BINDv9.  I wouldn't be at all surprised if 
they changed some key features of the coding model (specifically 
involving mutex locking) to make it more scalable on a multiprocessor 
machine in the near future, so that it could easily scale to handle 
20,000 queries per second (or considerably more) on the right 
machine, without requiring excessive configuration assistance, etc....


	In particular, Rick Jones is doing a lot of work on optimizing 
BINDv9 for HP, just as he previously did with BIND 8 (where he had 
gotten a single machine handling up to almost 14,000 queries per 
second, see 
<ftp://ftp.cup.hp.com/dist/networking/briefs/dns_server_results.txt>). 
He's turned up some interesting limitations inherent in the coding 
model that I would expect to be addressed in the near future.

	Indeed, looking back on the most recent preliminary results I 
have from him, on one particular machine he cranked BIND 8 up to just 
over 3660 queries per second, while BIND 9.1.0b3 on that same machine 
with the same test environment did almost 3300, which is almost 90% 
of the speed of BIND 8 but with an older version of BINDv9, and on 
another machine he got BIND 9.1.0 up to 10,000 queries per second. 
Rick's most recent report on BINDv9 tuning for HP/UX 11i is at 
<ftp://ftp.cup.hp.com/dist/networking/briefs/bind9_perf.txt>.

--
Brad Knowles, <brad.knowles at skynet.be>

/*     efdtt.c     Author:  Charles M. Hannum <root at ihack.net>             */
/*                                                                         */
/*     Thanks to Phil Carmody <fatphil at asdf.org> for additional tweaks.    */
/*                                                                         */
/*     Length:  434 bytes (excluding unnecessary newlines)                 */
/*                                                                         */
/*     Usage is:  cat title-key scrambled.vob | efdtt >clear.vob           */
/*     where title-key = "153 2 8 105 225" or other similar 5-byte key     */

#define m(i)(x[i]^s[i+84])<<
unsigned char x[5],y,s[2048];main(n){for(read(0,x,5);read(0,s,n=2048);write(1,s
,n))if(s[y=s[13]%8+20]/16%4==1){int i=m(1)17^256+m(0)8,k=m(2)0,j=m(4)17^m(3)9^k
*2-k%8^8,a=0,c=26;for(s[y]-=16;--c;j*=2)a=a*2^i&1,i=i/2^j&1<<24;for(j=127;++j<n
;c=c>y)c+=y=i^i/8^i>>4^i>>12,i=i>>8^y<<17,a^=a>>14,y=a^a*8^a<<6,a=a>>8^y<<9,k=s
[j],k="7Wo~'G_\216"[k&7]+2^"cr3sfw6v;*k+>/n."[k>>4]*2^k*257/8,s[j]=k^(k&k*2&34)
*6^c+~y;}}