2 posts • joined 10 Feb 2007
Abuse of standards process
The article fails to mention that Microsoft's distain for international standards got itself into this mess. A group of document format egghead users and programmers at OASIS spend years developing a workable standard. (BTW, the one of the criteria for that work was that Word and WordPerfect documents could be rendered with full fidelity.) They then submitted it to ISO, the ISO national bodies thought it was good and it was passed.
Some short time later, some governments concerned about archiving electronic documents decided to require new purchases to use the new international standard. That put Microsoft into a complete tizz -- we saw good people slandered, we saw backroom deals at their worst. But the power of the argument for using an international standard document format is strong, and very appealing to governments.
So Microsoft decided that it's own new document format needed to become an international standard. ECMA, who have evolved into a shop which can take your propietary format and make it an ISO standard, was called in. There has been a lot of disquite in the international standards field about ECMA's role, and Microsoft's behaviour is a perfect illustration of that concern. The behaviour is seen by most followers of the international standards process as an attempt by a threatened vendor to ride roughshod over a standard developed the way standards are meant to be developed. The international standards process has conciliation mechanisms to deal with this, but Microsoft failed to invoke those as ODF progressed to a ISO standard.
Why? Probably because Microsoft want OOXML as a standard *as it is*, not an altered compromise format. Which again comes back to Microsoft's distain for international standards.
OOXML doesn't even use international measurements of date and distance. You think ISO is pleased about proposed ISO standards specified in inches? Or that the leap year calculation is naively wrong?
There's also a lot of concern about the patent license status of OOXML. No one wants another RAMBUS. That's why OOXML is a much better abbreviation than OpenXML for Office Open XML.
Portraying this as IBM V Microsoft is wrong. Microsoft have upset way more many people than Big Blue: the good people at OASIS who made sure Word could use ODF; belivers in quality international standards at ISO and its national bodies; governments who want one international standard for archiving documents, not a vendor bun-fight; and lastly that 15% of people who don't use Microsoft's products and wish Microsoft would learn to play nicely with the other children.
That last group is important. Nothing impresses a national standards body more than a flood of individual letters from real people. You can see Microsoft futilely asking its faithful at various events to please, please write to their national body. To at least make it look like there is some support at the grass roots.
Gigabit ethernet needs stuffing with to be useful
Gigabit ethernet won't actually help much, although GbE does have the nice property of allowing a standard patch cable to be used to interlink the two interfaces (in fact a typical 10Base-T/100Base-TX crossover cable won't work for 1000Base-T).
The reason GbE isn't as useful as you'd hope is that the TCP buffers in Windows Xp are pitifully small at 12KB, which is adequate for dialup or a 10Base-T LAN but nothing faster or further away. When Xp came out they wanted to make it use as little kernel memory as possible, and they certainly wouldn't have been happy about the 4MB or more needed to keep a then-exotic GbE network interface card supplied with TCP data.
Pittsburgh Supercomputing Center have a good guide to tuning operating systems for network throughput. See http://www.psc.edu/networking/projects/tcptune/
Naturally the change on Microsoft operating systems varies by service pack and requires registry devilry and a reboot.