Hi Hemant, Sorry for using the term "inner header". I wasn't sure it would be unclear to anyone. Sorry again for not using the exact terminology as in the RFC.
That said it seems you miss the basic idea of load balancing. No, we certainly cannot use any set of fields, and this is very well known. The entire idea of using the inner header(TCP/ UDP) for load balancing is that we do not reorder micro flows. Reordering TCP micro-flows causes issues for TCP. Can you let me know any hardware which can use "any" set of fields? It would be interesting to know that. Thanks, Vishwas On Thu, Mar 20, 2008 at 6:48 AM, Hemant Singh (shemant) <[EMAIL PROTECTED]> wrote: > Vishwas, > > In one email you use the term "inner header fields" and in another email > you use the term "upper layer headers" to describe the same problem. The > discussion will not be fruitful unless we use a terminology that all of > us understand. > > Anyhow, that is one router implementation you speak of below in the load > balancing (LB) case. A router may use any set of fields to perform hash > for LB, not just the specific fields you speak of. The router LB is a > weak reason that is not strong enough for a case for this draft. > > Anyhow, I would like folks to wait till the Meeting minutes from 6man @ > IETF 71 are published to the mailer. You will see what the meeting > feedback on this draft was. > > I will reply to Tatuya in a short while. > > Thanks. > > Hemant > > > -----Original Message----- > From: Vishwas Manral [mailto:[EMAIL PROTECTED] > > Sent: Wednesday, March 19, 2008 1:23 PM > To: Hemant Singh (shemant) > > Cc: Suresh Krishnan; IETF IPv6 Mailing List > Subject: Re: Review comments for draft-krishnan-ipv6-exthdr-01.txt > > > > Hi Hemant, > > > <hs>Of course, firewall vendors will be biased towards inspecting an > > EH that is not the HBH. But that is not what RFC 2460 says. Here is > > text from RFC 2460 that clearly says, no intermediate node will > > inspect/process any EH besides the HBH. > > > > [With one exception, extension headers are not examined or processed > > by any node along a packet's delivery path, until the packet > reaches > > the node (or each of the set of nodes, in the case of multicast) > > identified in the Destination Address field of the IPv6 header.] > > > > Pardon my ignorance, but I need to see an explicit RFC that says > > firewalls being intermediate nodes, are allowed to inspect an EH that > > is not HBH and that RFC should also say that the RFC updates RFC > 2460. > > </hs> > It is not just firewalls, like I have mentioned earlier. In case there > are two routes to a destination which are of Equal cost (ECMP), a router > actually does load balancing by calculating a hash based on the inner > header fields like 5-tuple fields. I think this is basic to just about > any router and not firewalls itself. > > Thanks, > Vishwas > -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
