This is highly tangential to dnsop in general and this draft in
particular, but perhaps some clarity on the specific load balancing
point is useful.
On 3-Oct-2007, at 1233, Dean Anderson wrote:
On Wed, 3 Oct 2007, Brian Dickson wrote:
Load balancers can load balance, without any translation being
done at all.
Note "can" above.
And a load balancer is by definition doing *anycast*. The same
address
is used as a destination, and the packets are delivered to multiple
hosts. That is anycast.
No, that isn't
(always)
anycast. A loadbalancer is actually
(sometimes)
a stateful NAT with
several different hosts behind the load balancing NAT. Those
loadbalancer devices you buy from cisco and other companies are
(sometimes)
specialized NAT boxes
(or can be configured that way)
. The servers behind the load balancer have
all have different ip addresses
(but it's often possible to bind a single address to multiple origin
servers, e.g. as a loopback, and have the host process inbound
traffic which has not been subject to address translation)
As a data point, f5's BigIP supports load balancing of singly-
addressed services amongst many origin servers without address
translation, along the lines parenthesised above. Indeed, in many
cases this approach is recommended in order to maximise performance.
I have no recent hands-on experience of other vendors' load
balancers, but I would be surprised if f5 is unique in this respect.
Just one more tiny, procedural point: when you said
Then GROW considers an Anycast Draft, by your company.
that's not accurate. That document was co-authored by me when I was
variously an employee of ISC, a self-employed consultant to various
people, and an employee of Afilias Canada. The affiliation shown on
the document, as with all RFCs, is for the purposes of identification
only. I have never shared an employer with the other co-author.
Also, s/considers/considered/. The document was published some time
ago as RFC 4786/BCP 126. This is all old news; perhaps it's time to
move on.
Joe
_______________________________________________
DNSOP mailing list
DNSOP@ietf.org
https://www1.ietf.org/mailman/listinfo/dnsop
