Adam Tauno Williams wrote:
On Sat, 2009-10-24 at 19:53 +0200, Joachim Schrod wrote:
Adam Tauno Williams wrote:
2001::/16 - Allocated to RIRs 2002::/16 - Allocated to 6t4 fe80::/10 - Link local, those-MAC-derived addresses, at least for Ethernet. It uses a mechanism knows as SLAAC to come up with, at least on Ethernet, theoretically unique address. fec0::/10 - Site local, like 192.168.x.x, 10.x.x.x, etc... fc00::/8 - Unique Local, allocation still up in the air last I knew. But these are like a real network address but not routeable (?). fd00:/8 - Another kind of Unique Local, even more mysterious than fc00::/8. But you can get one from SixXS who seem to have appointed themselves as a registrar. This is what we use internally (one of these) for now. I.e., it is too difficult to plan and thus to set up in a small to medium company environment.
How so?
When you upgrade a service or router, crank up a new site, or replace some component you just make bringing up IPv6 part of the procedure. We just added it to our inspection / refresh cycle and in ~2 years we had IPv6 connectivity across the entire enterprise. We are hardly all-new-equipment [much of our production equipment is even EOL by the OEM] but it turns out *everything* except some printers, a couple DS1/DS3 banks, and a few IPMI interfaces supported IPv6.
I.e., you see this from the viewpoint of an IT professional. Most IT support personal in small to medium companies wouldn't even know what IPMI is. They will never have a DS3 connection to their company, much less a whole bank of it. This is also very visible from your other comments; i.e., that all companies that you have ever visited have Cisco or 3com network gear. You're looking at the problem from the very high-end of network deployments. (This is a factual statement, not an evaluation.) With DS1/DS3 banks, you might even work in a networking company. You are aware that this kind of work environment is very special and highly advanced, aren't you?
This is still a technology for big companies or trivial SOHO deployments.
Why?
1. SOHO IPv6 deployments work out of the box, no need to configure anything. 2. Big companies have a department with IT professionals that can spend the time to learn about IPv6, its advantages and its problems, and they might even be good enough to convince their management to spend money for the transition. As you wrote, IPv6 has actually very interesting features for complex and/or larger environments. And in companies of this size, there is (hopefully) a strategic approach towards IT: Here your argument that one runs IPv6 anyhow by plugging in new systems and one must control such deployments hits home. 3. Small- to medium-sized companies work under completely different premises. They often have an IT workforce of one or two persons who do application management, infrastructure management, system operations, network operations, and end user support. These guys (a) won't have the time to learn about address allocations like the one cited above, or other basic IPv6 principles. (And when they read the recommendation that they should run a daemon for such a simple thing as a default route, they are even less inclined to spend the time learning this "overengineered complex new stuff". After all, they have to keep their gear going.) And (b) the owner of such companies don't have an IT strategy where they are willing to spend money for work that does not bring short-term returns. Their IT strategy is "I have hired two guys and they shall keep my staff happy without asking for more money all the time". All your arguments in this thread are technical and are concerned with the 2nd use case. My argument has nothing to do with technology at all, and is all about the work dynamics of the 3rd use case. Joachim -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Joachim Schrod Email: jschrod@acm.org Roedermark, Germany -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org