On Thu, 4 Oct 2007, bill fumerola wrote:
> On Wed, Oct 03, 2007 at 08:10:03PM -0400, Dean Anderson wrote:
> > But none of this is relevant to the claims that Hickson made.
>
> no, but they're directly relevant to the claims that you made:
>
> >> direct server return aka one-arm load balancing does no translation or
> >> rewrite of any headers (l3 or l4). all it does is make a switching
> >> decision based on health check and other weighting criteria.
> >
> > Header rewriting? as in http header rewriting? That isn't anycast
> > either.
>
> you said a load balancers is "actually a stateful NAT". it's not.
And indeed, I agreed with you. You are correct: only some load
balancers are stateful NATs. There are indeed other kinds of load
balancers, as you pointed out. But none of them are implemented with
Anycast.
But by contrast, Dickson said:
"And a load balancer is by definition doing *anycast*."
which is still incorrect. By contrast, I admitted to correction in my
incomplete taxonomy of load balancers. Yet Dickson _still_ claims that
certain devices use Anycast in spite of evidence to the contrary.
It seems that you cannot admit to mistakes.
> as far as "no load balancers use anycast". many do route health injection
> which when you add more than one load balancer injecting the same route
> then you have....
BTW, (HSRP) Hot Swap Redundancy Protocol, and (VRRP) Virtual Router
Redundancy Protocol, aren't Anycast either.
But apparently you are defining stateful load balancers as "Anycast" and
then claiming success. That isn't success for Anycast, since it isn't
Anycasting. But F root, K root, etc are in fact doing Anycasting. They
are not using load balancers like the LTM to provide services.
> that thing you claim is a complete failure and can never work but
> somehow plenty of installations use it with success when configured
> properly within their environment.
I never asserted the term "Complete failure" for Anycast. I asserted
Anycast was __unreliable__ for stateful services. We don't have to show
"complete failure" to criticize the Root and TLD operations. We just
have to show their services are unreliable. Indeed, Anycast DNS services
are unreliable: I've shown in theory why Anycast is unreliable for
stateful services, and I've collected data that shows that the real
world matches the theory. Root and TLD operators and their crony's have
used hardball techniques to try to suppress that information. Next, I'd
like to collect NSID data on root and TLD anycast servers to have direct
data on root and tld operations.
> > And Hickson's dispute isn't relevant to anything about reflector
> > attacks. You apparently digress.
>
> i just must be a fraud and liar, not to mention a "junior sysadmin".
I try not to mistake simple ignorance for fraud and lies. But both
indeed deserve correction. A false statement becomes a lie after you
can no longer claim ignorance that the statement isn't true.
Scientific fraud comes in two types: fabricated data and fabricated
conclusions. If the data wasn't collected as reported, then the reported
data is fabricated. If the collected data doesn't imply the conclusion,
then the conclusion is fabricated.
Fraud, such as civil and criminal sort, has even more elements that are
necessary for a well-founded accusation. I won't go into those.
I don't expect you to meet scientific standards. However, I do expect
that you can reason and discover simple facts, such as whether the LTM
uses Anycast. I also expect you to be able to admit mistakes. I don't
think that is too much.
--Dean
--
Av8 Internet Prepared to pay a premium for better service?
www.av8.net faster, more reliable, better service
617 344 9000
_______________________________________________
DNSOP mailing list
DNSOP@ietf.org
https://www1.ietf.org/mailman/listinfo/dnsop
