Bruno Wolff III <[EMAIL PROTECTED]> writes on 9 August 2000 at 09:12:29 -0500
> On Wed, Aug 02, 2000 at 05:08:28PM +,
> JuanE <[EMAIL PROTECTED]> wrote:
> >
> > I did not think of that. Good suggestion.
> >
> > It seems like it would be a good compropmise if you can take your down
>
On Wed, Aug 02, 2000 at 05:08:28PM +,
JuanE <[EMAIL PROTECTED]> wrote:
>
> I did not think of that. Good suggestion.
>
> It seems like it would be a good compropmise if you can take your down
> server out of the rotation relatively quickly. If not, then you'll waste
> considerable time pol
Pavel Kankovsky <[EMAIL PROTECTED]> writes:
> On 2 Aug 2000, Frank D. Cringle wrote:
>
> > Generating a random permutation algorithmically is not too easy.
>
> Oh really?
>
> [ swap each element with a randomly chosen partner ]
Yes, that will do it. When I was originally working with this stu
On 2 Aug 2000, Frank D. Cringle wrote:
> Generating a random permutation algorithmically is not too easy.
Oh really?
int i, j, x;
int a[N];
for (i = 0; i < N; ++i)
a[i] = i;
for (i = N - 1; i > 0; --i) {
j = random(i);
x = a[i]; a[i] = a[j]; a[j] = x;
}
where random(i) i
Warning: This may be a stupid idea...
Could you take the number of milliseconds in the timeb stucture,
modulo , and use that as the first server to
start from? That could be done with a simple AND of a short,
and I think would provide a nice even spread across the servers.
And it doesn't re
: "JuanE" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Wednesday, August 02, 2000 12:37 PM
Subject: Re: updated load balancing qmail-qmqpc.c mods
>
> I agree with you both (Jay and Michael), at least partially. I agree that
> altough what Jay proposes will work, it i
e of use of down servers. Over time, if a server was
down for a long time, it would stop being polled for a connection as often.
- Original Message -
From: "David Dyer-Bennet" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Wednesday, August 02, 2000 1:03 PM
S
David Dyer-Bennet writes:
> JuanE <[EMAIL PROTECTED]> writes on 2 August 2000 at 16:37:36 GMT
> >
> > I agree with you both (Jay and Michael), at least partially. I agree that
> > altough what Jay proposes will work, it is too much computation and that a
> > simpler round-robin (after picki
JuanE <[EMAIL PROTECTED]> writes on 2 August 2000 at 16:37:36 GMT
>
> I agree with you both (Jay and Michael), at least partially. I agree that
> altough what Jay proposes will work, it is too much computation and that a
> simpler round-robin (after picking initial position) would suffice.
>
-Original Message-
> From: Michael T. Babcock [mailto:[EMAIL PROTECTED]]
> Sent: Wednesday, August 02, 2000 10:44 AM
> To: Austad, Jay; [EMAIL PROTECTED]
> Subject: Re: updated load balancing qmail-qmqpc.c mods
>
>
> Re-read my point: its unnecessary. I didn'
I agree with you, I forgot to mention that, sorry. I didn't have enough
Mountain Dew yet. :)
-Original Message-
From: Michael T. Babcock [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, August 02, 2000 10:44 AM
To: Austad, Jay; [EMAIL PROTECTED]
Subject: Re: updated load balancing
Re-read my point: its unnecessary. I didn't say it wouldn't work. I said
the CPU use of doing it this way was unnecessary over a simpler round-robin
approach (After picking an initial random server).
Note: I think using an array of pointers to server addresses would allow you
to do your rotatio
c(size);
p=array;
for(f=0;fmailto:[EMAIL PROTECTED]]
Sent: Wednesday, August 02, 2000 8:58 AM
To: JuanE; [EMAIL PROTECTED]
Subject: Re: updated load balancing qmail-qmqpc.c mods
Again, its unnecessary and does not add anything to the 'system' of
distributing load across servers.
PROTECTED]>
> <[EMAIL PROTECTED]>
> <[EMAIL PROTECTED]>
> In-Reply-To: <[EMAIL PROTECTED]>
> From: "JuanE" <[EMAIL PROTECTED]>
> To: James Raftery <[EMAIL PROTECTED]>
> Subject: Re: updated load balancing qmail-qmqpc.c mo
Again, its unnecessary and does not add anything to the 'system' of
distributing load across servers. Distributing in a true round-robin
fashion is sufficient if the servers are of equal quality (and/or drop
connections sooner if they are not). True round-robin distribution cannot
easily be acco
On Wed, Aug 02, 2000 at 11:13:22AM +, JuanE wrote:
> This way, you will redistribute the load randomly to all
> servers and not just the next one on the list.
Not quite. If choose truly randomly the liklehood of you choosing the
same server /every/ time is the same as the liklihood of choosin
I like the idea of statically listing them, but as you mentioned it is not
very scalable. The shuffling can be very expensive. I believe it would grow
with N^2. Instead of shuffling, just plain random sampling would be better
computationally. See me previous posting.
I think this is a "profile,
the array by a random number, it
> would shuffle it it randomly. That would keep it from banging on the server
> immediately after the dead one, and give everything an even chance again.
>
> Jay
>
> -Original Message-
> From: JuanE [mailto:[EMAIL PROTECTED]]
>
"Austad, Jay" <[EMAIL PROTECTED]> writes:
> From: JuanE [mailto:[EMAIL PROTECTED]]
>> Assume you have 4 servers A, B, C, D, and that server C is down. When your
>> random seed hits server A, B or D (with probability 1/4, respectively) the
>> message will go through fine. When your random seeds hit
hance again.
Jay
-Original Message-
From: JuanE [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, August 01, 2000 6:57 PM
To: [EMAIL PROTECTED]
Subject: Re: updated load balancing qmail-qmqpc.c mods
Jay,
If I understand this correctly, then I think there is a flaw in this type
of random r
Jay,
If I understand this correctly, then I think there is a flaw in this type
of random round robin. I think I can best describe it with an example.
Assume you have 4 servers A, B, C, D, and that server C is down. When your
random seed hits server A, B or D (with probability 1/4, respectively)
21 matches
Mail list logo