Friday Jun 08, 2007

Deep-Linking Into The Pink Dots

The "pink dots" maps have been getting some special attention lately. First, Jonathan referenced our original Solaris registrations map to make a point about how Sun's embrace of free software is driving adoption of our technologies. Then yesterday, Eduardo mentioned our new GlassFish adoption map on The Aquarium.

Both Jonathan and Eduardo asked readers to look at a particular map view to get an illustration of their point. One thing worth mentioning is that it's actually possible to link directly to such views.

For example, Jonathan (who wanted his readers to look at the map with a blank background) could have referenced this URL:

Or Eduardo (who wanted users to look at how GlassFish usage had increased in Brazil) might have referenced this URL for February:

...and this one for April:

But wait, those are complicated URLs. How could anyone possibly know which one to use?

It's easy. Just find the map view that you want and then copy the link referenced by the "This View" anchor in the map page. It's the link which I've highlighted in yellow below:

The JavaScript code in the maps page dynamically updates this link to always reference a URL which would recreate the current view.

So there you have it. If you want to have people look at some specific view of these maps to illustrate your point, you can send them there with just one click.

However, there is one caveat...

These deep-linking URLs guarantee that everyone will see the same map view at a given center point and zoom level. However, the actual amount of territory visible in the map (and thus summarized in the sidebar stats) will depend on the user's window size and screen resolution. So they won't necessarily see exactly the same image and figures that you do (though it should be close, assuming that most people have reasonably-sized screens and windows).

Wednesday Feb 14, 2007

Re: Is Solaris the New GNU?

Payton Byrd is wondering whether Solaris will become the new GNU. Specifically, his question seems to be whether OpenSolaris (if re-released under the GPLv3) could end up providing the preferred kernel for the GNU operating system.

It's a good opening question, but his argument then veers off with a lot of specific points which don't make a whole lot of sense. Let's look at a few examples.

Byrd: "Does the discontent this creates among Sun's engineers further push away the people who have made Solaris such an incredible product?"

What discontent? I think it's safe to say that most Solaris engineers at Sun are supportive of open source efforts and expanding the Solaris community. If adopting a more widely-accepted open source license advances those goals, I don't see why it would create discontent in our engineers.

Byrd: "If the FSF and Sun move forward with a replacement of Linux with Solaris in the GNU Operating System, I forsee a very ugly, protracted, and devastating fight that will last for years and seriously impede the progress that Linux is making into the market place. Whether this is a good or bad thing is a matter of perspective. I do know one thing, it highlights the fact that the GPL is anticompetitive because GPL v3 is looking to not only lock out IP protecting Novell, but Linux as well."

What you're calling a long and ugly fight, I would call ongoing and healthy competition. Has the existence of the BSD projects been devastating to Linux? Or vice versa? Has PostgreSQL been devastating to MySQL? Or vice versa? No. Competition is good. Where's the downside in having some competition and choice for kernels in GNU distributions. For some, Linux will continue to be the right choice. For others, a Solaris kernel may make more sense. And for both kernels' development communities, the competition and exchange of ideas will be a good thing.

As far as the "GPL is anticompetitive" statement... It makes so little sense to me that I wouldn't even know where to begin in responding.

Byrd: "I say let Sun, the FSF, and GNU fade away into oblivion!"

Huh? Earlier you say that Sun's Solaris is "probably the best kernel in the world." You correctly gave credit to GNU and the FSF for providing a huge portion of the software that people think of as "Linux". And now you're calling for all of the above to fade into oblivion, presumably because you've decided they're some kind of threat to Linux? Wow. Talk about being anticompetitive.

Tuesday Jan 02, 2007

Drilling Down in the Solaris Registrations Map

Recently Jim Grisanzio referenced our Solaris Registrations Map to illustrate the volume of OpenSolaris activity in Japan. Since I'd love to see more people using our map in this way, I thought I'd talk about it a bit.

For this kind of use, one handy feature is the ability to reference a particular map view directly. This can be done by copying the "Link To: ... This View" URL when you have the map zoomed and positioned as desired. For example, Jim could have referenced this URL when talking about Solaris-related activity in Japan. This allows everyone to look at approximately the same map view and statistics. (Note: the URL controls the map type, zoom level, and center coordinates; it obviously cannot control clients' screen resolution, which is what determines how much of the map around the center is shown and leads me to use the "approximately" qualifier.)

On my screen, this view currently shows:

Registrations In Visible Area
Solaris 10 / sparc:1715
Solaris 10 / x86:6387
OpenSolaris / sparc:17
OpenSolaris / x86:290

How do we interpret these numbers? Well, one thing they do not mean is that there are just 8409 Solaris users in Japan. As the FAQ notes, this map only shows data for "Solaris 10 and Open Solaris instances that activated Sun Connection to receive automatic software updates." As with any product, only a subset of total users will go through a registration/activation process.

I'm not aware of a good way to estimate what percentage of total users will have registered. So I can't infer the total number of Solaris and OpenSolaris users in Japan from this map view. On the other hand, it seems likely that whatever percentage of users choose to register/activate in one region would roughly equal the percentage of users who do so in other regions. If that's true, we should be able to use this map to compare the relative size of Solaris and OpenSolaris users in different geographic regions. For example, a fully zoomed-out view of the map currently shows a total of 83268 activated instances. So comparing our Japan total (8409 instances) to this number, we could estimate that around 10% of worldwide Solaris and OpenSolaris users are in Japan. That's interesting (and makes me think that perhaps we should update the map to show such percentages automatically).

Hopefully this gives you some ideas on how to dig for interesting views and stats in the map. As we've seen, the statistics it shows are not good indicators of the number of users or installations in absolute terms, but may be useful in estimating the relative populations for different geographic areas. We plan to investigate adding new data sets (such as Jim suggests) which may provide more absolute population info. Stay tuned. And let me know if you have ideas for a data set you think should be included.

Wednesday Dec 20, 2006

The Story Behind The Solaris Registrations Map

I'm one of the people who created the Solaris Registrations Map which has caught Jonathan's eye. Developing the map has been a lot of fun, and I thought I'd share some highlights here.

First off, credit where credit is due... The original idea for the map came from Steve Wilson, Vice President of the "SysNet" group at Sun. Among other things, this group runs the Sun Update Connection and the related Solaris registration service from which the map's data comes. Steve had only been on the job a few days when at small group outing he thought aloud: wouldn't it be cool if we made a Google Maps mashup of all of these registrations we're getting? As another of the map's developers would later say, to some of us Steve's question was the equivalent of chumming the water for sharks. We knew a great idea when we heard one, and we just had to make it happen.

So over the next few days and weeks, we started playing with visualizing our registration data using the Google Maps API. A few challenges quickly became apparent. Chief among these were performance and privacy concerns.

On the performance front, we quickly realized that the JavaScript-based markers often used for such mashups just would not work for a large data set such as this one (80,000 registrations and counting). So for our high-level views, we use the Java Image I/O API to create our own matrix of tile images, and then use Google Maps API calls to have those overlaid atop the normal map tiles. For example, where a fully zoomed-out view of the earth would cripple one's browser with the work of placing 80,000 individual "balloons" on a map using JavaScript-based markers, the custom tile approach means the browser just has to retrieve and position at most a handful images. Much faster and kinder to your browser.

Next, the privacy concerns... We had originally been assuming this map would only be available to a restricted audience on Sun's internal network. But soon after we started showing off initial versions, certain people were asking why we couldn't expose it on the Internet. Again, we knew a great idea when we heard one and had to make it happen. After all, Sun has jumped into the spirit of transparency and the Participation Age like no other. Making our map public could be one more step in that path, but it had to be done with care: we absolutely could not violate the privacy expectations of our users. As cool as some people might think it would be to find a dot on their exact house, others would understandably be quite upset.

And so, after consulting with Sun's privacy specialists we settled on an approach of locating registrations by nothing more specific than a zip or postal code. In other words, your dot won't show up directly on your house (unless, perhaps, you happen to live at a central post office). Instead, information from you and any of your Solaris-using nieghbors will be aggregated together and put into one marker at the center of your postal code area. As our team's Director, Eric Peterson, put it: even "Ted Nugent would be comfortable knowing that the T1000/Nevada registered to his 'Tedquarters' in Jackson, MI is untracable beyond 49202."

I think there are some interesting details deeper down in our technical approach, but I'll cover those in a future post. In the meantime, it looks like Steve has posted his own account of the map's story.




« February 2017