Friday, February 26, 2010

links for 2010-02-26: Killing innovation; KLM and Google Apps; Smartphones vs Notebooks

Lysacek vs Plushenko: A look at the numbers and data

In the wake of Evan Lysacek's gold medal upset in the mens figure skating at the Vancouver Olympics, silver medalist and pre-event favorite Evgeni Plushenko came out and said it: He was robbed.  His contention is that mens skating needs quadruple jumps, and that he did one, Lysacek didn't, and thus he should win.
"We need quadruples ... that is the future of figure skating," Plushenko said. "Without the quadruple, [it's] not men's figure skating. Now it's dancing."
The scoring system in figure skating has changed and is no longer the subjective 6 point scoring system where you had the mysterious "Russian judge" giving scores one couldn't understand, but is now a much more objective, particularly on the technical side, system where each element has a base score and the competitor is judged on how well they performed it.  Further, jumps performed late in the program earn a 10% bonus to give credit for performing them when more difficult and to encourage skaters to not front load their routines making them somewhat boring.  Add this all up and you have the technical score.

Now, a quad jump is indeed given a higher base score and so Plushenko's claim does have some merit, but it is only 1 of 13 elements that are scored, so since I love looking at data, I couldn't help diving in to the detailed scoring of their two routines to see how they stacked up.

In the table below, I've listed the programs for both skaters with the base value for their elements, their grade of execution, and running total along with the difference of their running totals through the programs.  I've also put in bold those jumps that occurred in the second half of the programs and received a 10% bonus for the base value.  Note, since both skaters had identical program components scores, analyzing the technical score is all that is needed to compare them.

LysacekBaseGoETotalPlushenkoBaseGoETotalLysacek Lead

From this we can see that Plushenko did take an early lead with his quad/triple combination and extended it with his early triple axel combination and triple loop, but then the lead began to shrink and ultimately became a lead for Lysacek when he threw his triple/double/double combination.  The way the lead varies is really due to the order of the elements as we aren't comparing apples to apples in the rows.  To do that, I've reordered Plushenko's program to match the rows of similar elements of Lysacek's below and the Diff column is the difference for that row/element.  The 10% bonus items still appear in bold.


Now we can more clearly see how they compare.  Looking at just the jumps to start, as expected, Plushenko's quad does give him a big advantage, but Lysacek executed his triple axel better eating into nearly a 3rd of that advantage.  Plushenko got a bit back with his triple axel combination being better executed (even though Lysacek got the 10% bonus here), but the 10% bonus and better execution got Lysacek nearly a point for his triple loop.  They both did a third combination getting the 10% bonus but Lysacek threw in the extra double loop giving him a 0.7 advantage.  But the big difference was Lysacek throwing his triple lutz late in the program got him a 0.6 advantage that he extended to 1.4 with his execution.

In the end, for the jumps, Lysacek had a base value of 58.23 (3.23 from the 10% bonus) and 5.24 for execution for a total of 63.47 to Plushenko's 59.33 base (but only 1.53 from the 10% bonus) and 4.44 for execution for a total of 63.77.  Plushenko's big lead from the quad was nearly completely lost due to poorer execution (0.8 point diff) and not maximizing his 10% bonus (1.7 point diff).  Also, due to doing the quad toe loop, Plushenko wasn't able to fit a triple flip in his program which is more difficult than the triple salchow and triple loop that he did perform.

Lysacek also performed more difficult footwork and spins having a base value of 16.7 with 4.4 for execution to Plushenko's 15.7 and 3.24 for a difference of 2.16.  This results in, from a technical mark standpoint, Plushenko's 3.8 point base point advantage for his quad being lost and Lysacek actually beating him on the technical mark by 1.86.

Now, if Plushenko has simply paid more attention to putting elements where they get more bonus, say moving the triple axel (0.82 bonus) and triple lutz (0.6 bonus), the margin would have been narrowed to 0.44.  That is less than the difference from his poorer execution meaning that ultimately his winning or losing was due to how he arranged his program and his execution, particularly his triple axel where he had -0.36 grade of execution.

So from the data, it does not appear Plushenko was robbed, rather his assertion that the winner needs to try a quad is only one part of the story.

But what do you think?  Do you think Plushenko was robbed?

Thursday, February 25, 2010

links for 2010-02-25: Amazon SimpleDB; SOA and the Cloud; Dual Licensing

Wednesday, February 24, 2010

Blog entries from the past

I had posted quite a few entries on my blog at Sun, some of them actually interesting!  As I don't have access to that blog any more, in an effort to preserve the entries and provide some continuity and a single place to go for them, I've started importing some of them to this blog.

They should appear to be posted on the date they originally were so won't show up as new entries at the top of the blog, so for those that may be interested, here is a sampling of those I've imported:

Tuesday, February 23, 2010

links for 2010-02-23: Stuck with Oracle or IBM?; Amazon most trusted; Apple in the enterprise

Monday, February 22, 2010

Boeing's Future of Flight

I'm a bit embarrassed to say I'd never been until yesterday, but I finally visited Boeing's facility at Paine Field and went on the Future of Flight tour they have there.  And I have to say, it is impressive.

Boeing has been building a number of planes at the facility for years, starting with the original 747 in the late 1960's.  They presently build the 747-8, 767, 777, and 787 there.  Given that all of these planes are large, a large building is required to house the assembly lines for them.

The building was built originally for the 747 manufacturing and subsequently expanded and has some staggering numbers associated with it as it is the worlds largest building in the world by volume:
  • 472 million cubic feet.
  • 4.3 million square feet, enough room for 75(!) football fields and you could fit Disneyland inside and have 12 acres to spare.
  • There are 6 Tully's coffee shops inside, including the highest volume Tully's in the world.
  • There are 2.33 miles of pedestrian tunnels below the factory.  The "short" direction they have tunnels running 1/6 of a mile.
  • Boeing has 1,300 bicycles in the factory to help employees get around.
  • There are 1 million light bulbs in the factory.
 There are more astounding numbers, too many to list here.  And inside, to see how the assembly lines work and see 6 assembly lines, each long enough to hold 5-6 complete or in progress aircraft is amazing.  Unfortunately they don't allow cameras inside so I don't have pictures, but you can see a few on Boeing's site.

I did take pictures outside as it was a absolutely gorgeous day (it doesn't rain all winter in Seattle) and Paine Field is a fantastic location with views of the Cascades (including Mt. Baker and Mt. Rainier) and Olympics from the same spot.  I also managed to show several new planes on the flight line including some 787's and a new 777 for FedEx, and 2 of Boeing's Dream Lifters, which are used to fly in many of the already built components for the 787 so that the plane isn't so much built as it is assembled and can be done in only 3 days.  The Dream Lifters are the largest cargo planes in the world.

All in all, it was a great tour and I'm glad I went.  If you are interest in planes and flight and happen to be in the Seattle area, I highly recommend taking the tour.

Thursday, February 18, 2010

Oracle and Sun's Open Source Projects

When I wrote my Analysis of Oracle's Strategy for Sun's Middleware, I made mention of Oracle's somewhat vague statements about the future of several open-source communities and projects.  GlassFish is a special case and I covered it as such, but a couple key projects that did not have clear plans articulated were OpenESB and OpenSSO.

For OpenESB at least, and perhaps this is an indication of what will happen with OpenSSO, we do have more clarity from an e-mail sent on the users alias a couple days ago.  Thanks to Frank for sending it, and he highlights a few key things about OpenESB and GlassFish ESB:
  • Oracle does not intend to let the project disappear and will keep it on
  • Oracle has scaled down the investments in OpenESB from what they were at Sun.
  • There is a hope the community will step up to fill the void left by the reduced investment.
  • Existing GlassFish ESB customers will be supported and can buy more licenses, but there is no intent to sell licenses to new customers.
There is also an updated governance document that continues to require a Sun Contributor Agreement (SCA) to be signed to contribute and outlines Oracle's special role in the project as steward and reserves the right to:
  • Accept/reject committers
  • Veto the roadmap
  • Veto changes to the OpenESB website
  • Accept / reject changes to this governance document
So in effect, they will continue to provide the infrastructure for the project but also maintain strict control of the project.  The question then is how these rights will be exercised.  My guess is that as long as what is proposed/contributed is an evolution of what exists they will remain fairly hands off.  However, should anything that is there become competitive to or a threat to an Oracle product, I'd expect the rights above to be exercised to some degree to address that.

It is also very important to note that by continuing to require the SCA to contribute, Oracle maintains copyright to all the code and thus the right to use anything developed and contributed in Oracle commercial product under a non-open-source license.  I don't think folks should be scared away by this alone, but it is something to remember and not be surprised by should features or code from OpenESB show up in Fusion Middleware.

This is good news for several of the companies that have formed around the development of components and implementation using OpenESB like Logicoy, ForgeRock, Pymma, Imola, Adjoovo, and more, as it means they don't have to create their own fork(s) and can continue to collaborate together in the community.

It is now up to the community to step up and contribute and see how this plays out for real, but the opportunity for them to do so is great news.  Will this same model be applied to OpenSSO?  We'll have to wait and see.

links for 2010-02-18: Oracle's Cloud; Windows vs Mac; JBoss AS 6 M2

Wednesday, February 17, 2010

links for 2010-02-17: NY Times and iPad; Facebook Harvested

Monday, February 15, 2010

Analysis of Oracle's strategy for Sun's Middleware

Oracle has done an admirable job of quickly communicating strategy and plans for how they will integrate, support, and invest in Sun's products and lines of business.  See their Transforming the IT Industry event and the Product Strategy Webcast Series for what they are saying publicly.  But their presentations only go so deep and there are many details left to interpretation.

So, we have analysis and opinions expressed by analysts like Coté and Stephen O'Grady at RedMonk and John Rymer from Forrester.  But a host of others have voiced their opinions and concerns and folks from Oracle have responded to some (see end of this post for a sampling) so I figured I'd weigh in with my view of what will happen, specifically to Sun's middleware products.  As I really like the Q&A format that @sogrady uses, I'll use it here too.

Q: Before we begin, anything to disclose?
A: Yes, while at Sun, at different times I ran Product Management for the SOA and Application Platform product lines as well as middleware and MySQL strategy.  I am no longer with Sun or Oracle now, and thus, everything written below is just my opinion and does not reflect any official position or statement from either company.

Q: For those that missed it, what happened with Oracle and Sun in the past few weeks?
A: The long awaited closing of Oracle's acquisition of Sun closed on January 27th, the same day as Oracle's high-level strategy event.  They subsequently released more detailed strategy webcasts on each of the acquired product areas.

Q: Cool, that's great that they did this, but can you summarize what is going to happen with Sun's middleware portfolio, starting with GlassFish fitting in with WebLogic?
A: Both Thomas Kurian and Hasan Rizvi clearly stated that WebLogic would be Oracle's strategic application server targeting enterprise deployments.  They also clearly said that GlassFish would continue as the Java EE Reference Implementation and an open-source project.

Q: Ok, that's great, but there is a lot of grey area in between.  Were any other details provided?
A: Yes, Hasan's webcast identified that GlassFish would continue to be support and sold as a standalone offering, but would also be added to all WebLogic offerings.

Q: So that's good for GlassFish then, right?
A: Mostly yes.  There was a mention that GlassFish would be targeted to departmental applications which would seem to indicate that they would maintain differentiation between GlassFish and WebLogic and push WebLogic for enterprise deployments.  However, in a response to a blog entry on the subject, Mike Lehmann (Oracle Product Management) hinted that some clustering and HA features that exist in GlassFish v2.1 would ultimately get into v3, so this appears to not be part of the gap.  There is still plenty to offer as differentiation beyond clustering and HA though.

The reason to have differentiation would be to not cannibalize WebLogic revenue.  They are already addressing this somewhat by including GlassFish in WebLogic offerings, but if it is available standalone they can't have a low priced GlassFish (pricing from Sun was a fraction of a WebLogic license) selling instead of WebLogic.  An alternative to having a lot of differentiation would be to raise the price of GlassFish to a level such that the lost revenue is small, or (and this is probably what they'd prefer, that a customer chooses to just pay a little more and get a WebLogic offering that now includes GlassFish.  It is this latter scenario that I expect to happen.  Of course, Oracle could choose to raise prices for WebLogic as well since it now includes GlassFish.

Q: Well, do they have to raise prices on GlassFish?
A: Not necessarily.  There was also mention during the webcasts particularly by Larry during his section that Sales was going to be comped on selling high margin items.  Thus, they might not raise GlassFish prices to WebLogic levels, but the margin on GlassFish would be lower so Sales would have less incentive to sell it.

Q: So is this good for app server customers?
A: It depends on the customer.  If they are an existing GlassFish customer, yes.  There is a commitment to continue developing and supporting the product for a long time.  Note that the development will likely stop short of providing features making it fully competitive with WebLogic though.  And at some point, renewals or maintenance will cost more than they do today.

If they are an existing or future WebLogic customer, yes.  WebLogic remains the strategic app server and should benefit through cross pollination with GlassFish as well as perhaps have Java EE 6 support accelerated.  Depending on how Oracle interprets existing WebLogic licenses, these customers may also get GlassFish for free since it is now part of WebLogic offerings.

If they are a future GlassFish customer, perhaps no.  I say that, not because GlassFish will suffer feature debt (although there will be some of that to maintain differentiation), rather it is because the price they will pay for GlassFish will almost certainly be significantly higher than it would have been from Sun.  Also, as noted above for existing GlassFish customers, some features that might have been desired may never make it and be reserved for WebLogic.

Q: Ok, enough about GlassFish, what about the rest of the Application Platform portfolio?
A: The Sun Web Server, likely rebranded/renamed to avoid confusion with the Oracle HTTP Server, will continue to be supported and sold perhaps as a peer to Oracle HTTP Server.  I expect that the Sun Web Stack will not be invested in and just existing customers supported.

Existing Sun Portal and Sun Web Space Server customers will be supported, but neither product will be invested in as Web Center is the strategic portal offering.

GlassFish MQ (OpenMQ) will stick around as part of GlassFish but won't be promoted on its own.

Q: Whew, that wasn't as simple as I'd hoped.  Is it simpler for SOA?
A: Generally yes.  Oracle did not identify any of Sun's SOA portfolio as strategic, but did pledge to support existing customers for a long period of time.  Further, they said that they would create bridge technology to allow collaboration between Java CAPS and Oracle SOA Suite, although this should effectively already exist through JMS and Web Services support today, so I'm interested to see what is added.  They also mentioned credits for customers wanting to convert from Java CAPS to Oracle SOA Suite.

There is less clarity or certainty around GlassFish ESB (and its open-source OpenESB project) as the statement simply said it would continue as an open-source project.  If the rumors on the community mailing list are true though, a large portion of the SOA staff were not kept which raises the question of who will be continuing the project?  Perhaps an opportunity for the community to step forward.

Oracle also said that the NHIN CONNECT open-source project would be supported.  Since portions of this was originally based on OpenESB, it is unclear how this will be accomplish though with a potentially short staffed group of engineers.  There is also some discussion on some mailing lists that NHIN CONNECT is really just using EJBs in GlassFish and not (that much of) GlassFish ESB.

Q: Didn't Sun have some cool MDM technology that came from SeeBeyond?
A: Yes, you do know this space don't you!  The roots of the technology are in healthcare and it is the Healthcare and Life Sciences vertical at Oracle that will be integrating this technology into their offerings according to that groups strategy webcast.

Q: Ok, how about identity and access management products?
A: The short answer is that Sun's Directory Server Enterprise Edition (DSEE) continues as a peer to Oracle's Internet Directory, both being strategic, Sun's Role Manager will be strategic, but the strategic products for the rest of the portfolio will be Oracle's existing products.

As far as OpenSSO goes, we get the amorphous statement about the open-source project continuing, but it was also clearly stated that tools would be built to help OpenSSO Enterprise customers move to Oracle's strategic products so don't count on OpenSSO being enhanced.

I did find it interesting that in order to provide clear naming, they are bringing back Waveset to the name for Sun's Identity Manager.  Waveset was the name of the company that Sun acquired around 6 years ago that had the product that became Sun Identity Manager.

Q: Last, a lot of Sun's middleware had integrated tooling in NetBeans.  What will happen to NetBeans and that tooling?
A: Yes, Java CAPS and GlassFish ESB bundled NetBeans and provided plugins for doing all SOA and ESB development.  Similarly, NetBeans 6.8 delivered full Java EE 6 support and integrated GlassFish in its downloads to provide a great developer offering.  Oracle clearly stated that JDeveloper is the strategic IDE for Fusion Middleware, they also have Eclipse based tooling (ironically just announcing a new release), and thus having a third full blown IDE in NetBeans probably isn't in the offing.  The comments seem to indicate that while NetBeans will be continued, its focus will be as a lightweight Java and scripting IDE.

Q: Didn't Sun have some Eclipse tooling too?
A: Yes, the GlassFish Tools Bundle for Eclipse provided a bundle with Eclipse, GlassFish, and plugins for Eclipse providing GlassFish, and in its most recent edition, some Java EE 6 support.  I'd say there is a good chance this gets rolled in to the Eclipse based tooling mentioned above as a way to accelerate getting Java EE 6 support in the offering.

Q: So what is the net-net for customers of Sun's products?
A: The good news is Oracle has promised to support the products for a long period of time, thus the investment is safe in that regard.  Oracle has a strong history of supporting acquired products and that should continue in this case.

The bad news for most of the products is that they are not likely to be significantly enhanced and getting important new capabilities will likely require migrating to Oracle's equivalent products.  The other bad news is that when it comes time to renew or do that migration, prices could very well be higher, either due to price increases or because Oracle may be more particular about counting servers/CPUs used and perhaps even perform audits.

Q: Anything else?
A: As noted above, these are just my opinions.  I might be wrong on some points, and Oracle can and will adjust their plans based on customer and community feedback, so I could end up being way off in the end.  If you want to prove me wrong and are a customer, let Oracle know what you think to get them to adjust their plans.  For other opinions, see some of the links below.

Sampling of other's opinions:

Friday, February 12, 2010

links for 2010-2-12: America's Cup; Apple the new Microsoft?

Thursday, February 11, 2010

links for 2010-02-11: Opera for iPhone; Buzz on Google Buzz

  • Opera Mini for iPhone - It will be interesting to see if pre-announcing it actually gets it through Apple's approval process.  I'm guessing no as Apple would say it confuses or negatively affects the users experience (i.e. competes with Safari).
  • Random thoughts about Google Buzz - 24 hours later - “Cant Google just BUY Twitter? And ditch the 140 character crap? Wouldn’t that make our lives easier?”

Wednesday, February 10, 2010

links for 2010-02-10: 15 suggestions for Oracle; WebSphere on Solaris; CIOs love free software

Tuesday, February 9, 2010

2010 Vancouver Winter Games - Argh, no CBC!

A week ago or so in anticipation of the start of the 2010 Vancouver Winter Olympics, I wrote a blog entry in which I expressed thanks that I (and those in Seattle or other close to Canada cities that get CBC) would be able to not constrained to the delayed and packaged for Joe American coverage that NBC provided as I'd have access to CBC's coverage which would likely be live and with more detail.  Alas, I was wrong!

I based this on the fact that CBC had covered the 2008 Beijing Summer Games splendidly and I found myself watching them live and just recording and perhaps watching later the NBC coverage.  Unfortunately, it appears that CBC did not win the rights for the Vancouver Olympics, instead CTV did.  Am I'm not the only one that made this assumption and am now disappointed.

So, I'll now be searching for ways to watch events live on the internet, or recording NBC's coverage to watch it at my own pace where I can skip all their fluff.  After all, if they are delaying it for me in the first place, what more harm is there going to be if it is delayed a bit more by me!

Note to NBC, this defeats the whole purpose of your advertising as I skip all the ads when I watch it off the DVR.  If you want me to watch the ads, show it live so I have reason to watch it live and endure the ads.

If anyone has any pointers to alternative ways to view the Olympics on the internet, please comment.

Monday, February 8, 2010

links for 2010-02-08: Oracle acquires AmberPoint; Super Bowl search trends

Sunday, February 7, 2010

links for 2010-02-07: Saints win Super Bowl; Ken Jacobs leaving Oracle

Saturday, February 6, 2010

My Farewell to Sun

I was notified a week ago that I would not be making the transition to Oracle and today was technically my termination date.  I came to Sun through the acquisition of SeeBeyond in 2005 and enjoyed the past 4.5 years working with fantastic engineers, creative marketeers, and just generally a bunch of great people.

I will miss many things at Sun from the culture, to the products (stay tuned for more thoughts on that in an upcoming blog entry), but most of all the people.  To those that are making the transition to Oracle, represent our products well and take advantage of the opportunity that working for Oracle provides.  To those that like me are not, best of luck in your future endeavors and let me know how I can help.

As far as I go, I will enjoy a little time off to see if I can improve my tennis and golf games and watch some Winter Olympics, but am definitely looking forward to the next challenge, whatever that may be, from middleware to cloud to development to who knows what.  My LinkedIn profile has a summary of what I've done and is a good way to contact me.

(image courtesy James Gosling)

Thursday, February 4, 2010

links for 2010-02-04: iPhone and Android; Kindle going after iPad?; Why Microsoft is 'faling'; AR parody

Wednesday, February 3, 2010

links for 2010-02-03: PHP Performance; Datacenter Pods; Apple Pricing

Tuesday, February 2, 2010

The Vancouver Olympics are 10 days away

The Winter Olympics in Vancouver, BC, Canada are only 10 days away!  It has been only 18 months since the Beijing Olympics, but thanks to the switch in the 90's to have the winter games during the "off" even years, we get to experience all that is the Olympics again.

Unfortunately, something that seems to come up every time the Olympics come around is the subject of drugs and doping.  The constant battle of the testers to catch up with the latest drug or masking technique continues with the cheaters usually being half to a full step ahead although someone usually gets caught and the testers can declare victory.  Given this, I was pleased to see that Italy will fine any athlete caught doping $140K!  All 109 athletes going to the games will have to sign a statement agreeing to pay the fine if caught.  From the story:
CONI president Giovanni Petrucci says the committee has added the deterrent "to show how serious Italian sport is in the fight against doping. Whoever doesn't sign won't go to Vancouver."
Other Olympic Committee's would be well served to follow Italy's example as a further deterent against doping.

But enough about the negative aspects of doping, what about the games and sports themselves?  In the United States, we'll get a steady diet of the glamor sports of figure skating, alpine skiing, speed skating, and hockey, but we'll get helpings of bobsledfreestyle skiing, and unfortunately snow boarding (how this became an Olympic sport I'll never know).  Interested in something else?  Good luck finding it on NBC, but that may actually be a good thing (see below).

For those of us living near the Canadian border, we are lucky enough to get CBC and additional coverage that way, but that could mean just adding curling to the list above!

For the US viewer, there will be a number of interesting stories including:
There will be many more stories I'm sure, and we can depend on NBC to show us as many of them as they show the sports in action.  Which brings us to the television coverage.

It appears that, as was done for Beijing and prior games, NBC will elect to show us plausibly live coverage of all the key events and perhaps more, instead of giving us live coverage.  Sure, a few events may align with the eastern time zone to be shown live there, and the lower tier events NBC doesn't really care about may be live on secondary channels, but much of the coverage won't be live and worse, the events that are live will only be live in the eastern time zone and those of us on the west coast only a few hours from the events themselves, will have to wait 3 hours for all that coverage whether taped or live.

In this day and age of, Twitter, e-mail, Facebook, etc. it is just silly for NBC to think that folks can really hide from the results and thus the drama they are trying to hype will be lost with folks knowing the results.  That alone may cause some to not tune in, but also with so many having DVRs/Tivo/etc. now, if NBC isn't showing it live in the first place, many (I know I will) will simply record NBC's coverage and watch it at our own speed.  Then we can skip the ads and avoid the "up close and personals" which also defeats the purpose of NBC trying to get everything into prime time so they can sell those ads.  Perhaps those of us doing that are in the minority and NBC is really appealing to the average TV viewer who will tune in.

For a longer and more eloquent rant on the subject, see Bruce Jenkins article.

Thankfully, in addition to having a DVR, I will have access to CBC and I expect while they may not have as many total hours of coverage that NBC claims, they will show more events live and I plan to take advantage of that.

What do you think?  What are you looking forward to from these Olympics?  Leave a comment!

Let the games begin!

links for 2010-02-02: Google Web Server; Flash required?; BEP/CEP

Monday, February 1, 2010

links for 2010-02-01: iPad ads and Flash; Personal Cloud; Nexus One and AT&T