MS volte face embraces next generation Internet
After six months of silence, are we going into IPv6 overdrive?
Yesterday Microsoft released a preview version of IPv6 - the Internet Protocol version 6, sometimes known as IPng. or next generation. But why Microsoft has suddenly decided to do this, in view of negative remarks it was making six months ago? Last June, Microsoft was a founder the IPV6 Forum - along with mostly network operators and equipment manufacturers - but the company made its policy towards IPv6 clear last August in a statement to Network World: "Microsoft has made significant investments in exploring this technology. However, due to the experimental nature of IPv6, Microsoft will not support it in Windows 2000, but will continue to solicit customer feedback and explore implementations in future versions of Windows." The claim of "experimental" is untrue of course, since v6 had been a stable draft-standard protocol of the IETF for some years, replacing the 20-year old Ipv4, which is increasingly running into problems as a result of growth. It was believed last summer that Internet pioneer Vint Cerf was lined-up to nudge Bill Gates on the desirability of Microsoft adopting v6, but it came to nothing. From many points of view, a move to IPv6 is desirable as quickly as possible, but the player controlling most clients - Microsoft - didn't want to play. Indeed, there seems to have been a split in the Microsoft camp between Microsoft Research (which appeared to welcome participating in the v6 development work) and the Windows development team, which probably saw it as just another problem that it would rather forget about while it debugged Windows 2000. Consequently, Microsoft Research seems to have acted like a maverick, releasing its stack with the source code and producing experimental versions of IPv6 for NT 4.0, as well as the Windows 2000 beta. Yesterday, Microsoft made its grudging step and accepted that "an explosion in the number of users and devices connected to the Internet, combined with projections for the future, mandates the need" for IPv6. The lame excuse was that the tools were now available for the transition, Y2K worries had gone away, and that this was the time to push Windows 2000 by making an "IPv6 Technology Preview" available. But it "does not work on any version of Windows NT, Windows 95, Windows 98, or Windows CE", and it is "not officially supported" by Microsoft. Trumpet Software of Tasmania has developed an IPv6 protocol stack for Windows 9x and NT, but its effort was unsupported by Microsoft. Microsoft also stated that the Preview "should not be deployed in a production environment". The eventual migration to v6 "will be a gradual one", Microsoft said, without explaining why it was unnecessary to work at Web speed, as it had been doing with XML development. Could this be a bluff, we wondered: would Microsoft now be secretly working flat-out in an endeavour to move ahead of the present state of the art, introducing modifications that would only work with Microsoft software? Surely not. But why was it necessary that only Microsoft's Visual C++ could be used? Was it just a coincidence that the Preview requires that "the Internet Protocol (TCP/IP) supplied with Windows 2000 must be installed", and that "if you currently have an Ipv6 protocol installed, such as the Microsoft Research IPv6 protocol, you must uninstall it" first. In itself, the use of the word "uninstall" is pretty interesting, in view of Microsoft's claims about the un-uninstallability of key, integrated parts of its operating systems. But with no black-hat wearing Netscape types around, the guys in the white hats could perhaps afford to be more liberal. So far as support is concerned, "Microsoft does not provide any level of technical support", but Microsoft had graciously set up a public newsgroup for feedback, to help it with the bugs. There is some comfort however: in the End-User Licence Agreement, "Microsoft's entire liability, and your exclusive remedy under the EULA shall not exceed five dollars." But let's return to why Microsoft has decided to put a toe in the v6 water. Perhaps Microsoft has at last woken up to the realisation that there soon will be billions of WAP devices and other mass-market Internet devices needing to be linked to the Internet, and that it had better do something about controlling the market in the way it knows best - by proprietarising the standard. There's an alternative and perhaps more attractive hypothesis. Could it be that Microsoft's sudden enthusiasm for IPv6 results from its inability to deal with denial of service attacks and design limitations, so far as security is concerned? IPv6 gives the possibility of solving quite a few security problems at a network level, and could remove the need for Microsoft firemen to move from blaze to blaze, scattering urgent bug fixes. If so, we can at least take comfort from the certainty that these firemen would not lose their jobs in the switch to v6: there are many more fires smouldering away in Windowland. ® Related story: What the hell is - IPv6?