Sunday, April 24, 2005

Solaris 10 + dual-core Opterons = record breaking performance!

For many years now, I've been a SPARC bigot. The only machines I've owned (except for really old 8-bit stuff), until recently, were SPARC based (modulo my 68030-based Sun 3/80). But for all of SPARC's virtues, there's no denying that x86's (and especially AMD's) price/performance is fantastic, especially when mated to a high performance OS like . Sun recently released the latest benchmark results for their 4-way, dual-core AMD Opteron server, the V40z. This combination smashed the previous record; more details in the press release.

If this doesn't convince the nay-sayers that the "Slowaris" moniker no longer applies, nothing will.

Friday, April 08, 2005

Rebutting some anti-OpenSolaris FUD

I really have to respond to this piece of uninformed claptrap. It is almost 100% pure FUD. Frankly, I can think of only one reason why extreme Linux zealots like Maureen O'Gara write stuff like this: they are scared shitless of the competition Linux is facing with OpenSolaris. Deep down, they know they can't win on technical arguments, so they resort to spreading as much FUD as possible.

I'll quote sections of the article, and then respond with the facts.

This time through, people are pretty convinced Sun "doesn't get" open source - or doesn't want to. Murmuring is heard about Sun's stated intent to cherry pick any community changes made in the name of "OpenSolaris" and put them in a Sun-blessed, Sun-tested, Sun-supported, Sun-distributed Solaris "subset."


How can Sun cherry pick a source base that is open? It is true that Sun will continue to release its own branded version of Solaris, in much the same way that StarOffice is the Sun-branded version of OpenOffice. But if anything, the Sun-branded Solaris will be a superset of OpenSolaris, including those bits that can't be opened because the IP holder didn't grant Sun the rights to open the code. Again, this is no different to StarOffice and OpenOffice.

Sun has yet to post the OpenSolaris code though it was promised weeks ago. And it's not going to be Solaris 10. It's going to be Solaris 10.1, widgetry Sun has code named Nevada. How they differ is unclear.


Ummm, Sun promised no such thing. They have committed to releasing the code in Q2 this year; by my reckoning they have nearly 3 months left, given that Q2 started 8 days ago. And the reason why it won't be the source code to Solaris 10 is quite simple: the code that is released is based on the current version under development. The code that makes up Solaris 10 FCS is about 4 months old now. As for the comment about how they differ being unclear: perhaps if Maureen did just a little research, she could read the What's New notes on the Solaris Express page or read one of several blogs detailing the changes.

The stuff is now due out by the end of the quarter under Sun's new Open Source Initiative-approved royalty-free Common Development and Distribution License (CDDL), a variant of the Mozilla License that forbids putting Solaris code in Linux or connecting it to any GPL code to avoid having to put proprietary IP in the public domain.


Sigh. I thought this one was addressed months ago: the CDDL is not incompatible with the GPL; it is the GPL that is incompatible with the CDDL. It is perfectly permissible to link code under license with CDDL code, provided the license of the other code allows it. GPL code can only be linked to other GPL code; it is the GPL that is inflexible and restricts developers' choices, not the CDDL. Oh, and releasing IP under the GPL does not place it into the public domain. Something in the public domain has no copyright and can be used as one sees fit. The GPL has some very explicit provisions copyrights and what one may do with the code.

Within hours of its CAB announcement, Sun president and COO Jonathan Schwartz, Scott McNeil's stand-in as Peck's Bad Boy, slammed the GPL as predatory economic imperialism while keynoting the Open Source Business Conference and said Sun would remain aloof from it.


Who's "Scott McNeil"? I presume Maureen means Scott McNealy, Sun's CEO. And it would be hard for Sun to remain aloof from the GPL, given that is the license used by OpenOffice...

Sun also stuck two of its own people on the CAB: chief technology evangelist Simon Phipps and senior staff engineer Casper Dik. Hmmm, Phipps.


Not sure what Maureen means by these thinly disguised attacks. Both Casper and Simon are respected, consumate professionals, as are the rest of the CAB. (In the interests of full disclosure, I should point out that I am a member of the CAB.)

The bottom line is this: we in the community encourage and enjoy healthy competition. Competition fosters innovation, and that's what it's really all about. If hard-line Linux zealots like Maureen put their energies into making Linux a better product rather than indulging in pointless FUD spreading, their OS of choice might have a better chance of keeping up...

The real real CAB

So the cat is finally out of the bag--I am one of the two people (Al Hopper being the other) voted by the OpenSolaris pilot project members to represent the community on Sun's newly formed OpenSolaris Community Advisory Board (CAB). (Unlike last Friday's April Fool, this really is kosher!) I am honoured to be part of the CAB, and I thank those in the community who placed their trust in me. The pilot program was notified of the election results several weeks ago, and the identity of the other CAB members was also revealed as soon as possible. The public announcement was timed to coincide with the OSBC in San Francisoc, and Sun flew the five of us (i.e., me, Al, Casper Dik, Simon Phipps, and Roy Fielding) into town for our inaugral meeting and to do our bit (more on this anon). I am looking forward to working with these talented people, doing our bit to guide Sun's opening of the source code.

Our first official activities took place on Monday morning, so we all arrived sometime on Sunday. Casper and I were the first to arrive (he was checking in just as I walked in the Crowne Plaza hotel's lobby), and we had planned to meet the rest of the gang for dinner at around 17:00 (5:00pm). After freshening up, I met Casper in the lobby, who was playing with his Ferrari 3400. I commented on this, saying that I'd also brought mine. Then he pointed to the screen and said "ah, but I bet you don't have a WiFi driver on your Ferrari yet". (This was true, and do I really need to point out that we were talking about Solaris drivers? Running Windoze is not an option, or at least, is to be avoided if at all possible.) I replied that I didn't, so he showed me some other cool stuff: the improved mouse pad driver, and the AMD PowerNow driver. Those following the Solaris on Ferrari story will recognise that I'm refering to Casper's almost-mythical "frkit". After some pleading and whining on my part (and an unspoken promise not to redistribute said goodies), he acquiesed the next day and gave me a copy of the elusive tar ball. Actually, I think it was his indignation that I needed to use "some other OS" for these features that finally convinced him! :-) Suffice to say that I am posting this via a WiFi connection from my Ferrari 3400 running Solaris 10 and frkit.

Anyway, back to the story. Al Hopper was the next to arrive; although none of us (with the exception of Casper and I) had ever met before, only exchanging email, Al was a dead give away: he was wearing the T-shirt he got from the Solaris 10 launch! Our fearless leader, Jim Grisanzio, arrived next, with Sara Dornsife from marketing. Finally, my buddy and fellow pilot program member, Ben Rockwood arrived. We talked for an hour so, getting to know each other, waiting for Simon and Roy. Eventually hunger got the better of us, so we went for a meal in the hotel's restaurant. After dinner (chocolate dessert for me of course!), we talked for a few more hours until we had to go to bed through sheer exhaustion.

The next morning we went over the road for breakfast, in a cafe whose interior reminded Jim and I of the one from the scene in Pulp Fiction where John Travolta's character takes his boss' wife for dinner at a 50's-themed restaurant, where they share a $5 shake, and win a prize in the twist competition. Here, we finally met Simon and Roy. After food, it was off to Sun's downtown office for our inaugral meeting. I was the last of our mob into the elevator and just behind me, a jean-wearing ponytailed man slipped in. "Hey," I thought to myself, "he looks like Jonathan Schwartz. Wait, it is Jonathan!" We exchanged hi's, and that was that (for then). I'll not bore you with the details of the meeting (Al and I brought our Ferraris, and had them "tuned" by Casper) as the important bits have been reported elsewhere, but we were due to have a 30 minute chat with Jonathan. 90 minutes (and several sandwiches) later, he was still there! Meeting him was one of the highs of the day for me; he's a great guy, and you know he is fully commited to this OpenSolaris thing when he says after introductions "so, what can I do?". We told him the main issues we'd identified up to that point, and he pledged to make the problems go away.

After some more discussion, it was time for our 15:00 press conference call. Slated to last only 30 minutes, this too went on longer than planned due to the interest of the journalists and analysts on the call. I think we had about 30 participants on the call (excluding ourselves), quite a high figure the PR people present later told us.

After a bit more work, we walked to a restaurant called Lulu's Place. It was here that we met with yet more press, after which we had dinner with numerous Sun engineers and executives. Pardon the name dropping, but the Sun people included (in no particular order, and no slight intended for those that I miss--I have a terrible memory for names at the best of times!): Andy Tucker and Bryan Cantrill (both of whom I've had the pleasure of meeting before), Mike Shapiro, Dan Price, Glenn Weinberg, Stephen Harpster, Stephen Hahn, and Tom Goguen (there are at least two VPs just in that lot...), not to mention Sara Dornsife, Claire Giordano, and Terri Molini. All in, there was about 25 of us, and a very good time was had by all.

We stayed very late, and finally had to go because we had an early start the next day.

The next day we met in the foyer of the hotel hosting the OSBC, registered, and then listened to Jonathan's keynote speach. Afterwards, we grabbed a bite to eat and then headed off for the Sun room, where several hours of press interviews awaited us.

While we were waiting for the evening reception to start, I had an interesting conversation with one of the vendors. They were selling these cool little servers, about the size of a laptop power supply. I think the spec included 64 MB of RAM and a dual core PPC processor, running Linux 2.6. "What would be really cool," says I in full evangelist swing, "is if that thing was running OpenSolaris". The vendor's techie and I had a respectful and engaging conversation about the pros and cons of our respective OSes of choice, but after I pointed out S10's innovations like DTrace, zones, privilege separation, etc., etc., he was forced to concede that unlike many claims to the contrary, Linux's unstable ABI is not necessary for innovation.

Moving on, the reception was a good chance to meet with lots of people, including one Microsoft ASP.NET Program Manager who was on the receiving end of some of my pent-up frustration of Windoze and Microsoft's illegally gained monopoly. At around 21:00, everyone had had their fill of Solaris brand wine(!), so accompanied by Sara, the five of us and Ben Rockwood went out for a bite to eat. It was Sara who observed that I "really do have a chocolate problem"; not being very hungary, I ordered just a dessert called Chocolate IV. Oh, incidentally, I scored Jonathan's conference name badge from his executive assistant, Noel! :-)

On Wednesday it was time to pack and go back to Edmonton. I had hoped to see a few more sights, like Lombard Street, but I guess that will have to wait until the next time I'm in the City by the Sea...

Friday, April 01, 2005

Will the real CAB please stand up?

Just in case anyone still believes it, my last posting, about the members of the CAB, was an April Fool! (That being said, I do know who the real CAB members are...)

Congrats to those of you who spotted it, and bonus points for anyone who sussed that "Avril Dummkopf" is a mixture of French and German for April Fool. :-)

OpenSolaris Community Advisory Board (CAB) members announced!

As many of you know, Sun is planning to release the source code to Solaris under an OSI-approved license, called the CDDL. To aid them in this process, they have announced that a Community Advisory Board (CAB) will be put in place. The CAB will initially have five members: two nominated from the OpenSolaris pilot project, two Sun employees, and a fifth person who is prominent in the Open Source community, appointed by Sun. Although it was formed a few weeks ago, the members of the CAB have so far remained anonymous.

Although Jonathan Schwartz will officially announce the identities of the CAB members soon, Sun's OpenSolaris Director, Avril Dummkopf, has given me permission to post their names here (I am a member of the OpenSolaris pilot, and have been since day one).

With no further ado, the CAB are:

  • Marshall Kirk McKusick (OpenSolaris pilot member)
  • Eric S Raymond (OpenSolaris pilot member)
  • Andy Tucker (Sun employee)
  • Bryan Cantrill (Sun employee)
  • Linus Torvolds (Linux's head honcho)


Marshall is of course one of the chief architects of the BSD OS, and ESR is well known by many for writing the New Hacker's Dictionary, and The Cathedral and the Bazaar. The appointment of Andy and Bryan, being senior members of the kernel team, comes as no surprise.

I emailed Linus about him accepting a position on the CAB; apparently he did it because he wanted to see what Solaris was really like, having recently been quoted as branding it "a joke". My guess is that he also wanted to see which bits would be easiest to "borrow". RMS (Richard Stallman) was also in line for the fifth position, but apparently he felt his time would be better spent taking singing lessons...

Congrats to the new CAB, and I for one look forward to seeing the fruits of your labours!