RossoWorld

This site and various others that I host from home, have been for the past 4+ years running on a Sun Netra X1 server. This, at the time, was a sweet little 1U rack mountable server priced under $1000 with a Sparc IIe 500Mhz processor, 128Mb of Ram and a 40GB drive. It was all I needed to run a web site without having to put a system together myself.

The only problem I found with it during this time for home computer use was the noise. Server class machines are never quiet and most are actually very loud. If you’ve ever stepped into a computer room you know what I’m talking about. The cooling requirements for reliable operations are such that high speed and noisy fans have to be used most of the time. This one was barable at 6 bels but even behind closed doors of my closet it was still audible.

In addition to this, I’ve been running more and more software on it for storing pictures of our daugther and running some backend processes for the winery. With that little RAM the machine was constantly swapping.

So I’ve finally decided to take the plunge and upgrade. This time I was looking for something that didn’t require any huge and loud fans, and that would put out little heat (since the systems I have hang out in the closet).

I saw that Intel had recently come out with their new laptop chips and I thought I’d give that a look. I finally decided on the new Intel Core Duo. It’s a dual core, low power chip used in the latest laptops. I picked up a 2Ghz version of it for a good price on Ebay and found a micro-ATX motherboard from ASUS that would actually support it. That along with a fast SATA II drive, 1GB of DDR2 Ram, a quiet power supply, and a 2U case gave me what I want…. performance and (almost) total silence. The system is incredibly quiet and runs really cool. Even after having it on for a week the CPU heatsink isn’t even hot to the touch (unlike most idling CPUs).

I ended up installing Fedora Core 5 on it and using most of the packages provided right out of the box… Apache 2.2, Tomcat, etc. I had to downgrade to PHP 4 from PHP 5 because of some older software I’m running, but the compile time for PHP 4 was incredibly fast, taking a few minutes rather than 10-15 on the old system.

I’m sad to see the Netra go. I really wanted to upgrade to one of the new T1000 that Sun has just come out with but they are too expensive (although relatively cheap in server terms) and even louder at 7 bels. Oh, well. I was tempted to install Solaris 10 x86 on the new box but I finally went with FC5 partly to try it out and partly because of all the various software that come pre-installed. I might try Solaris 10 yet, especially to try out the virtual ‘zones’ it has to run mutliple installation on the same box.

Performance now is blazing fast. PHP scripts run instantaneously, and mysql queries are in the milliseconds.

I’m a bit worried about the relability of this new server. As I write this, the Netra has been up for 477 days without a reboot or failure. I don’t know if I’ll be able to get this kind of reliability with hardware that I put together myself and that hasn’t been tested all together.

We’ll see how it goes. Anyone need a Netra X1?

Lately I’ve been refocused towards Operations rather than Engineering (hence blogging less about the ArcWeb APIs than I would like to) and taking a hard look at our Infrastructure for ArcWeb and hosting.

I had the opportunity to attend HostingCon 2006 about a month ago where I was able to see how hosters manage their users and the technologies they use to share hardware resources between the thousands of users they have.

One of the latest buzz words (or acronoyms in this case) in the industry is SaaS. This stands very simply for Software as a Service. Citied examples at this conference included software such as CRMs, Mail, etc. Here is the wikipedia definition:

Software as a Service (SaaS) is a model of software delivery where the software company provides maintenance, daily technical operation, and support for the software provided to their client. SaaS is a model of software delivery rather than a market segment; software can be delivered using this method to any market segment including home consumers, small business, medium and large business.

The term might be new but the concept has been around since the internet came into its own. Hosting applications or exposing software as web services is something that has been done for quite a while. Appliances are another incarnation of SaaS. If you take a look at ESRI’s offering of ArcWeb, ArcData Online, Geography Network, etc. these could all be considered SaaS. SalesForce.com is another. In the past software wasn’t really designed to be hosted and served as a service, so wrappers had to be developed. Today, software has shifted to being re-developed to be ‘internet friendly’ or even be ‘internet centric’.

Take Turbo Tax for instance. From the looks of it each year, more and more of it is being written based on browser technology: HTML, JavaScript, etc. Why would that be? Writting HTML is much easier today because of the tools and technology than to write your own custom forms, plus, Intuit can easily offer the software as a service (which they do).

Also take a look at the evolution of Internet GIS at ESRI. Back in the days, you could take ArcView 3 and expose it on the web using ArcWeb IMS. It was clunky because you were running a desktop product to feed functionality on the web. Now with ArcIMS and ArcGIS Server which were designed to live on the internet, GIS can be really easily provided as a service.

There will always be reasons for a company to use software in-house themselves but as the technology gets more functional it also becomes more complex, and takes more to learn, install and manage for the customer. As internet pipes get bigger and bigger and networks gets faster and faster, latency between client and server will be practically nothing. This means that it won’t make any difference for a customer to be using local software which needs to be installed, configured, maintained, secured, etc., vs. a hosted version of the software which you can just log in and use. It makes a lot of sense for a company, especially if they are not savy in a particular area, to outsource the task to someone who makes it their core business.

Even looking at it from the the hosting company’s perspective SaaS makes sense. Instead of investing a lot of time in installers to support all types of DBs, Web Servers, Applications Servers, Operating Systems, etc. out there (not to mention technical support), they can now focus on the core functionality, performance, security, and everything that makes the product what it is. With the cost of hardware going down and large multi-core/multi-processor machines becoming much more prevalent, it becomes really cost effective to provide software as a service.

Take a look at the current ESRI software stack as an example. For a saavy GIS user using something like ArcSDE + ArcGIS + ArcIMS might be fairly straight forward. But think about the non-GIS users who have nothing GIS related in their enterprise and who know little about GIS. They might really benefit from a GIS but setting up a complete GIS takes people to run it, learn it, configure it, tune it. Many companies shy away from this and go with sometimes simpler but less functional solutions, thus cutting them out of a lot of the power and flexibility they could be getting.

Now think about how easy it would be if someone ran all this for you, and all you had to do is access it across the internet. That would be something…

Well, that’s ArcWeb for you. Take a look at something like Business Analyst Online which runs on the ArcWeb infrastructure. You could buy the book (or DVD) of source data for all the 11,000 variables on our servers, but then you would have to build the entire system yourself. Why? It’s not your data. All you want to do is use it to figure out if a particular area is suitable for your store or what kind of customers are around you to focus your marketing, etc. So log in and just use it!

Same goes for the ArcWeb Services developer APIs themselves. Why would you want to maintain street basemap data from yourself? Just use the functionlaity across the web (of course, unless you maintain a much more detailed street database) and build it into your application.

I really believe that this is where the software industry is going. This is basically your turn-key system or appliance, bundleing the software you need along with the hardware, internet connection, and everything else you would need.

There was one remark at the closing session of the ESRI UC which caught my attention above other. A user asked if it would be possible to make some of the features in the Enterprise ArcGIS package available online on a on-demand basis. This way a customer can pay less for their base package and then possibly use the more advanced functionality on-demand. This is a great idea whose business model needs furthur exploration, but it indicates that users have gotten the message. On-demand, SaaS, or whatever we call it is here to stay and useful. Pay only for what you need.

At ESRI we’ve had the concept of Managed Services for a while now, under which we are hosting various larger sites for customers. The idea of building an application based on ArcWeb and then having ESRI host it is very appealing to many, as it means that there is not hardware infrastructure to manage on the customer’s side.

Managed Services is now graduating to the next level. Up to now it has been for specific projects that came to us through various channels. Now we want to open this up so that it is much easier for everyone to participate if they’d like and need to. Hosting your ArcWeb (or other ESRI based) applications and providing SaaS back to our customers.

The User Conference was a good discovery mechanism for us on this topic. There were a few sessions on our Managed Services offering which had a fairly good turn-out and there was quite a bit of traffic at the ArcWeb Services island talking about this. I think we got some good feedback which we’ll use to improve our offering.

I think in the distant future we’ll get to the point where home and work PCs are simply terminals onto the internet and we will simply switch on the computer, connect, and use all our software on other servers. ArcWeb is a step towards that vision of SaaS. It let’s the techies focus on what they like best and everyone else just enjoy the product of their efforts. For me, being a techie, this is a challenge and an opportunity to contribute to the latest trends of software technology. Let’s see how it plays out.

The 2nd Annual ArcWeb User Group Meeting will be held at the User Conference on Wed. Aug 9th, from 5:30pm-7:00pm on the East Terrace Mezzanine Level.

Last year was the first user group meeting and it was a success. We had a good showing of users and had some great food and giveaways.

This year with the latest releases of the SOAP APIs, REST API, JavaScript API, etc., there’s even more to talk about and find out. So drop by. Most of the ArcWeb team attending the conference will be there so this is your chance to give them feedback and get the latest news on ArcWeb.

See you there.

Ok, I finally had a little time to setup and migrate the site to WordPress. I’m still tinkering around with the theme (for now it’s a default one), and have to get a few things working again. In general all the posts are there, the old links have been mapped to the new ones, and the RSS feed should be redirected to the new one. I didn’t migrate the comments (as I didn’t have too many).

If I missed some old post, or some link is not working anymore, let me know.

Fantastico.

Italy went all the way and for the 4th time in the history of the World Cup have won it again. I was a bit disappointed in the final as their performance wasn’t really all that great in the second half and over time but I’m still glad they ended up winning it all. The penalty shot at the 6th minute for France really made my hopes sink but the Italians were able to equalize soon there after. France really dominated for the most of the second part but didn’t manage to score against the all star Italian defenders.

Things got a bit interesting when Zidane (the ‘pillar’ of the French team) for some reason hit one of the Italian players with his head while play was stopped. I was glad to see he was thrown out for doing this. I’m surprised to see this from a veteran player who was playing his last national game and especially during a tied game.

The game reached the shoot outs and for the first time as far as I can remember the Italian did not miss a single one! I was dreading a final like in ‘94 when they missed (lots of them). Penalties are never a satisfying way to end the game but with players exhausted the play would never end.

I still remember when the Italians won in ‘82. I was in Italy then and right after the match ended just about every car in my small town was honking their horns, people outside cheering, etc. Quite the celebration. I’m sure it’s the same there now, but here in Redlands, nothing. Oh, well.

Let’s hope for the best for the Italian players in South Africa 2010.

I’m not much of a sports fan for the most part except for when it comes to Formula 1 racing (go Ferrari!) and World Cup Soccer (forza Italia!).

This year the world cup is shaping up for a great finale. For the first time since ‘82 it’s an all European semi-final. Tuesday and Wednesday will see Germany vs. Italy and Portugal vs. France. My wife being German we’ll have quite the fireworks on July 4th. With all the talk of game fixing and the possibility of some of the best teams in Italy being demoted to the smaller series, the Italians didn’t start out in their best form… but with luck and some great talent they’ve gained quite some momentum and now the cup is within reach. Germany supposedly never has beaten Italy in a World Cup match but the Germans have been playing very well. I think we’ll see quite the battle on Tuesday.

So even for those who aren’t soccer fans, I hope you’ll watch the semis (and the final next Sunday), it’s going to be quite a show.

Forza Italia.

(Oh, and the Formula 1 season is shaping up pretty well too with Ferrari catching up to Renault with their double win at Indianapolis).

I had the opportunity to attend once again the JavaOne conference this year. I quite enjoy from time to time the ‘full immersion’ you get from going to such a conference. The Java Community has grown a tremendous amount since the first JavaOne I attended (I think in 1997 or 1998) and there is so much good stuff going on it’s hard to keep track. The conference is very much like our ESRI User Conference with its variety and interesting topics and it is a great place to catch up on all the latest technologies and developments.

Here’s a summary of the conference… First, let’s get the bad stuff out of the way. I was really irritated at how the whole session attendance was done. This year you had to sign up for each session you wanted to attend. This was so that they could gauge how big a room they needed and if they should repeat the session again. Maybe a good idea in theory but in practice they had to scan the badges of every single person going into a room and this caused massive lines to form, some even going outside the conference center. Waiting in line is not my favorite pass time… Years ago when they weren’t scanning badges I found that the crowd control was much better even when the conference had 15k attendees.

Then, there was the bad wireless connection. The whole conference center was pretty much covered and you got a pretty decent signal all over but the performance of the whole network was really really bad. The connection (not the wireless signal) would drop every so often, and each time you had to open your browser to get logged in which was extremely annoying… I think the cause of it was just the massive amount of wireless devices in the hall (probably every single person there had something) but you’d think you would plan for that especially for a conference where mobile, wireless, etc. is very important.

Other than that the conference was very good. This year the sessions were actually very good. In some previous years there were a lot of intro sessions which barely got into the details of the topics. This year it seemed like presenters got into much more of the guts of things with lots of code and detailed explanations.

AJAX was probably the #1 buzz word at the conference. Not so much in plenary presentations but more from the tool vendors, the sessions, etc. Any session with the word AJAX in it was packed. AJAX is cool but in the end there doesn’t seem to be that much to the technology so I was rather surprised. Several tool vendors have built components to facilitate AJAX interaction. Toolkits like DOJO and Prototype which hide the AJAX communication for you were talked about quite a bit.

Sun announced lots of new OpenSource projects. They open sourced their AppServer last year (GlassFish) and have quite a community around it now. They are building their latest version of their commercial app server (which we use) on top of the Open Source version. They also open sourced their JMS queue and have announced that they will be open sourcing Java which is quite big news.

There were quite a few sessions on REST. Now there is support for REST in JAX-WS and they were showing how to use this. Interestingly when asked ‘When should I use SOAP vs. REST’ the presenter of one of the sessions I went to responded that REST is generally used when re-publishing content from web sites such as Amazon, news Sites, etc. My biggest gripe about REST at this point is that there is no WSDL equivalent and so it makes it much harder to deal with all the different ways people expose REST out there. If you’re trying to integrate lots of different web services it seems like SOAP is much cleaner at this point. It sounds like there are various projects working on this issue. We’ll see where this goes…

Along with REST and SOAP, there were some interesting talks on SOA, which was a big topic of the conference, with ESB architectures, integration of various systems, etc. The new Java EE 5 version (which is out) also has some very interesting features for SOA built in. Things like just putting a @WebService as an annotation to a class will expose it as a web service auto magically. Lots of stuff is taken care of for you and with the work they are doing with .NET compatibility it should be pretty seamless in different environments. Looking forward to start using it.

And of course the fun part of the conference was the slot car racing contest. This was a demo of Real Time Java which allows you to control the motor of the slot car and sense when you are over a given sensor. It was an interesting problem since you only really knew when you were over a sensor and how much time elapsed from there, so you had to be tricky on how you managed the voltage to make the car brake (which sometimes resulted in it going backwards) and accelerated around the curves. I spent a little time trying it out and managed to get around the track in 27 some seconds but the fast guys were down in the 25 some seconds (of course they had spent their whole conference working on this). Prize for winning was a Sun Workstation.

Wireless was still at the forefront of presentations with the sale of the latest SAVAJE OS phone which could run Swing, Java 2D, etc. on it. Pretty much any Java SE app could potentially run on the phone as opposed to special Java ME apps. I considered getting one to play with (they were fully unlocked GSM phones for $200) but then again I probably wouldn’t have the time.

There were some very interesting new software for management and deployment of apps which I’ll be looking into when I get back. The infrastructure for ArcWeb is getting quite large and various tools could help us be more efficient when doing deployments and managing all our apps.

Overall, it was a good conference, and it always helps to spark ideas and creativity in our team after talking to all the ‘experts’ in their fields.

The last wine tasting at Annette’s was canceled because of unforeseen events but it has been rescheduled to tonight. Same deal as before. I think I’ve e-mailed most of you that might be interested but if you want to try the new 2004 Amelie come on by and give it a try.

After trying some of the latest bottles of 2004 Amelie, I think it’s more than ready to go out to the public. So late last week we officially released it on the Mankas Hills Vineyards web site. It’s a real nice blend of Cabernet and Merlot. Give it a try if you get a chance.

At the same time I also finally had a chance to put up a slightly re-designed version of the web site. I really wanted to have the rounded corner box look like several other sites and it turned out to be a lot more work then I thought! I had some issues with laying things out with tables and getting the right height and found a few sites talking about table-less layouts using just CSS. I managed to get it all implemented but it really took a while to trouble shoot between browesers and I finally had to resort to a bit of JavaScript to get IE to look right. Although CSS and JavaScript are relatively simple the amount of time spent in debugging and troubleshooting in different browsers is just insane. I hope IE 7 and later versions of Firefox at least will play in the same way in the future.