[Linux-aus] My take on the Microsoft "Competitive Hour" (long)

[Leon Brooks]

> The presentations were professional but disappointing.

Agree.

> The overall tone was "us versus them" which will sadden
> the heart of most networking staff, who are at the front
> line of making Windows and Unix interoperate, and who
> want Microsoft to play nicely (as opposed to sly alterations
> to Kerberos, a AD/LDAP structure which tries to force AD
> to be the core directory when SAP or PoepleSoft makes
> more sense, ever-changing document formats, an unwillingness
> to write decent HTML 4.0 that a XML-aware web server could
> get its teeth into, etc).

Admittedly, they did (in the Perth session at least) state up front that the 
presentation was biassed. Call it either chutzpah or courage, but it was 
there. (-:

The presentation was not well attended (only about 20 people) probably because 
it was scheduled - by accident or design, I know not - directly opposite the 
Windows 2003 presentation.

> The Linux presentation relied upon the recent IDC study into
> Windows and Linux TCO, the one where an IDC anaylst later
> said Windows came out best because Microsoft knew what
> questions to ask to get the answers it wanted.

A customer of mine had asked a few questions about that IDC study, starting 
with `why are the Linux licensing cost nine-hundred-odd US dollars?' and 
including `why is the manpower considered equivalent?'. I passed those 
questions on, mid-presentation, pointing out that they were from a customer, 
alluded to the customer not believing the study and consequently not 
believing Microsoft, and that I didn't know how to answer them. Mike's 
response was to hypothesise that maybe support costs were included in that 
(asked if support costs were included in Microsoft's $211, Mike said yes, 
must take him to task for that one day) and finally settled for `we'll get 
back to you on that', so I said, `OK, but is it published anywhere? How will 
the other people here get the answers?' And eventually they promised to 
publish those answers in their partner magazine. Stay tuned!

>  - Listed competing products as: MySQL, OpenLDAP, Samba,
>    Apache, Linux.  [No mention of PHP.]

They did mention PHP in ours, as a web scripting language - three letters, no 
explanation. They also leaned heavily on AD vs LDAP, stating (correctly) that 
LDAP represented only a small part of what AD could do and implying (falsely) 
that there was nothing to replace the other parts.

There was a ? on one entry of their Linux table, and a ? at the bottom. They 
promised to put the presentations on their partner site, so later I might be 
able to tell you what the ? represents. They didn't often present any 
alternates in the FOSS column, like PostgreSQL, Tux, *BSD, Kerberos etc, but 
were generous in so doing in their column.

They also made much of FOSS copying what Microsoft did. I was being a good boy 
and chewing heavily on my tongue, so did not point out that accomodating 
everyone else was trivial, Microsoft were the only serious player who made 
things so difficult that a separate project was needed to cope with the 
damage.

>  - Mentioned competitors as IBM and Sun. [No mention of
>    Red Hat.]

Or anyone else. Much harping on Linux only as an escape from expensive 
proprietary Unices. They wish. (-:

>  - Most Linux TCO studies were flawed, the cost savings
>    mainly coming from moving RISC-->Intel.

There _are_ those USD$1k Linux licence fees to contend with, after all! (-:

>  - Building a Microsoft community -- online, certification,
>    Most Valued Professional, TechNet Plus.  6m developers,
>    1.4m certified people [they avoided the work "engineer"],
>    14,000 certified trainers, 2,200 user groups.

Skipped most of that with us, did mention 25,000 people employed in their 
engineering division (wondered how many of those were secretaries, gardeners, 
IT support, internal couriers etc).

>  - UNIX application migration.  New product "Microsoft
>    services for UNIX" to allow simple porting of POSIX
>    applications.

Neglected to mention that heart and soul of SFU is no less than gcc...

>    Mainframe portrayed as a expensive 1Ghz
>    CPU [which misses the point, mainframes are an expensive 1GHz
>    CPU which *never stops*].

Also, a 1GHz CPU which is finely tuned against memory and IO, completely 
compartmentalisable, everything hot pluggable (-: even the Linux scheduler is 
hot-pluggable now, Tony B-T tells me :-), stacks of redundancy and 
parallelism, obscene amounts of storage, extremely clever storage controllers 
etc. Should have asked him how much a single decent SCSI RAID controller 
costs even for a crappy PC.

>  - Pushed SQL Server as a fast, cheap-transaction database.
>    [ Obviously written before SQL Slammer, the cost of which
>    which made the TCO argument of a few percentage points
>    against Oracle a bit silly. ]

Also obviously written before Linux had more than 4 TPC entries. I'm pretty 
sure (see, for example, http://www.newsfactor.com/perl/story/20890.html - now 
might be Microsoft's very last chance to claim cheaper TPC). There's another 
interesting but not updated since 2001 benchmarking commentary on this page:
http://www.kegel.com/nt-linux-benchmarks.html

>  - MS have a single vision -- .NET on Windows.  Linux is
>    all over the place [one of those places perhaps being
>    where you would like to be].

Also harping on their expectation that Microsoft will always be around, and 
will always exhibit firm leadership. Does anyone remember how narrowly 
Windows 95 missed being shipped without Internet capability? Or any of the 
other score or more of leadership volte-faces within Microsoft? And as to 
always being around, well... http://www.billparish.com/

>  - Security.  Windows 2003 locked down by default.

A Windows-using mate of mine immediately asked them to add a `suicide button', 
meaning, `I don't care, I want everything switched on'. (-:

>  - 30% of documents import incorrectly into StarOffice.

Ours claimed 1 in 3 with `content or formatting' issues. Private discussion 
they admitted that one in 20 wqas more reasonable but claimed that Gartner 
had tested documents from all levels of enterprise (maybe referring to a 
Gartner study: http://gartner2002.hec.ca/research/107800/107883/107883.html). 
And would ya look at the flaws in that! Ongoing project management, for 
example, swamps the cost of everything else. It does at least admit that 
going from MS-Office to MS-Office is not cost-free.

Interesting stats:

  * `number of users: 1750';
  * `documents requiring conversion per user: 25';
  * `...at the enterprise: 2000' - eh? Can they do basic maths?
    25*1750 == 43,750 documents;
  * `average number of macros per user: 0.05';
  * `% existing PCs re-imaged: 5%' - in relation to SO - what for?
  * `% existing PCs migrated via network: 0' - why not?

> Macros don't import at all.

They changed this to `many macros don't import'.

> Password protected files don't import [wrong].

But true for their version of SO 6.0. It prompts for the password, then fails 
to open the file.

> Some silliness importing a spreadsheet and looking
>    and the #REF errors [to my mind this was a good thing -- you
>    got errors rather than a bogus result.

Their point was not the #REF errors per se, but that they were due to a failed 
macro (true, but overblown).

> In the real world,
>    I don't find the Word import errors to be significant, as the
>    result is usually a better-looking document.

They also failed to treat the case of moving from OOo to MSO, probably because 
it would then quickly become obvious that MSO was the product that was being 
difficult to deal with.

I and my customers also use OOo for repairing and porting documents that 
either kill MSO or won't port between MSO versions (which would you rather 
have, the odd formatting quirk or sudden-death application failure).

> By far the
>    majority of spreadsheets import just fine.  PowerPoint imports
>    are the most problematic. ]

Odd, all I've ever lost PPT-SXI are fonts/symbols and fancy transitions.

>  - Some silliness cutting and pasting from a web browser into
>    Word and Writer.  Word was better.  [Not mentioned was the
>    difference is because Word imports the formatting whereas
>    Writer applies the current document's styles.  Beginners
>    would prefer the Word design choice, writers of bigger
>    documents would prefer the Writer design choice.  These
>    divergent views of which user is best catered to happens
>    throughout the two products.]

Yes. Another thing that struck me was that when SOW didn't paste the image, it 
said `Read Error: blah blah' (message crunched to fit in image bounds) and I 
wondered if perhaps IE had copied the image as an OLE2 object rather than as 
an image... perhaps worth experimenting. Does the same thing happen with 
Mozilla?

>  - StarOffice price is less than Office price.  But Gartner
>    says you are looking at $1,500 per seat to transition from
>    Office to StarOffice.  [There was no consideration of
>    new users -- which don't face transition costs -- at
>    schools or buyers of new computers or speakers of languages
>    not supported by Microsoft.]

I strongly suspect that those costs are drastically overblown, and ongoing 
costs of continuing with MSO like downtime or loss due to MSW or an MSO macro 
virus killing your whole machine or broadcasting your trade secrets across 
the 'net were ignored.

> I wouldn't be expecting
> a Word import filter for Writer documents anytime soon.

However, Michael Kleef assures me that the new MSO document format is 
plain-text-XML and is going to send me a sample to examine. Which of course 
means that OOo import/export filters for the new format won't be long coming. 
However, older versions of MSO will be completely unable to deal with the new 
documents, which to my mind looks like a golden opportunity for OOo.

Now... back to my own notes...

The first presewnter was Mike, and he hit Linux. He's the same Mike who did 
the SEAWA Java-vs-.NET presentation.

MS limited Linux to the following spaces: embedded, edge-of-network, 
high-performance (HPC) and Unix replacement. Anyone else notice the yawning 
Microsoft-shaped gap in that lineup? Wishful thinking or denial on their 
part. Linux started life on exactly the hardware that Microsoft's aiming to 
cover, so would of course be most effective on that hardware.

The corresponding reasons for this they listed as: a modular OS (Andy 
Tanenbaum's apoplexy on hearing that claim notwithstanding), traditional 
preferences (for an OS that MS are trying to portray as young, brash and 
inexperienced? Riiight... more dried frog pills, that man), HPC descending 
from Unix, and an easy-to-port escape from commercial Unix costs.

Their answers to these niches are `designer' versions fo their OSes (CE, XP 
Embedded), SOA, promoting their own HPC solutions, and a Unix-to-Windows 
migration guice (and SFU, which is GCC anyway, so why not undergird it with a 
real OS?). In summary it shows that they either have no clue what Linux is or 
how it developed, or are being wilfully ignorant for business' sake.

They made a big thing about integrating tools.

They promoted DRM, so you could distribute an internal memo via email and not 
have it broadcast over the entire net (unless someone took a photo of the 
screen while it was displayed and distributed that). I thought: `Halloween! 
that's their answer to the Halloween memos!' :-). To make DRM work as 
promised, you have to hermetically seal the entire computing environment, and 
there's a strong element of idealism in that which is going to be treated 
quite roughly at the hands of the real world.

The bias in the wording they used everywhere was incredible. If porting 
something from Unix to Linux was easy, they called it `quick and dirty'. If 
porting it to Windows, they ignored the difficulties and called it 
`integrating' the application (like, yeah, whacking it into a text window is 
total integration?).

Interestingly enough, they almost blipped right over interoperability, with 
the notable exception of importing their hairy data files into other people's 
software, and the claim was not `we made that difficult' but `their tools are 
no good at it'. <thwack>

But I digress. They claimed that Windows' security was comparable with the 
diverse Linux distributions, a cheat's way of saying that if you added up all 
of the vulnerabilities separately, the sum was almost as bad as Windows by 
itself. Render that down and you get: our security sucks, but by counting the 
same vulnerability once for every competitor, we can be made to look good.

They trundled out the obsolete and heavily slanted TPC benchmarks mentioned 
above, and claimed that Windows was cheaper per transaction than anything 
else, several times cheaper than Linux.

Microsoft are pushing integration (translation: stuff that's hard to produce a 
replacement for) and collaboration (which only works as presented if you 
presume a Palladium like MCP/BigBrother framework to set it in). They kept 
coming back to `the value in integration'.

Microsoft are ploughing $5G into stuff that's related to R&D this year.

They are also making their developers accountable for any piece of code they 
write or sign off as secure (hello, CVS tools?).

They assert that FOSS methods don't work for security, citing the recent 
sendmail vulnerability (aside: who uses sendmail any more?) as one found by a 
security company, not by FOSS developers. Er, yes? Said security company is a 
FOSS developer, and your point is...?

Steve, the antiStarOffice dude, started by asserting that he didn't want a 
feature by feature comparison, which he called "feature nitpicking", he 
wanted to sail on and compare the grand vision. And then later he showed the 
WMA video mentioned above. Ah, well, we should be used to that by now.

Steve - a Canadian - used to sell laser printers, and it shows.

Their sales technique is to turn the conversation from "costs" to "value" and 
then presumably trot out the bogus studies.

A strong point of MSO and Windows he emphasised is that they are `familiar' 
and `consistent' - he described customers expcting them as if they were `part 
of the woodwork'.

One curious term Steve used for FOSS was `free like a puppy' and used an 
example of a $20 rabbit that he agreed to his wife purchasing for the kids; 
when he got home, there were an indoor and outdoor hutch worth $500, food, 
and all kinds of other expenses not mentioned up front.

His point was that costing should be done inclusively, but I bet he didn't 
intend that this include costs of stuff like viruses and reimaging. (-:

Other than that and the comments above, the StarOffice presentation was pretty 
much content-free and heavy on visuals and emotion. A whole herd of one-trick 
ponies.

I was disappointed in Steve's lack of content. I'm looking forward to seeing 
Mike's presentation up on the Web so I can reconsider the details in it.

Cheers; Leon

-- 
http://cyberknights.com.au/     Modern tools; traditional dedication
http://plug.linux.org.au/       Committee Member, Perth Linux User Group
http://slpwa.asn.au/            Committee Member, Linux Professionals WA
http://linux.org.au/            Committee Member, Linux Australia