As Amazon continues to advance its cloud offerings for EC2 and reports strong financials, it seems like the only news I am hearing out of Google is about layoffs, cutting services, and yet unreleased cloud initiatives like GDrive and support for Java on AppEngine. I understand that AppEngine != EC2 but from a marketing perspective when you hear the word “cloud” you almost always hear Amazon or EC2 mentioned in the same sentence. Also, most of the success stories on cloud computing (Animoto, WashingtonPost.com, SmugMug, and Harvard Medical School) are all on EC2. Where are all the wildly successful AppEngine companies? I’ve played with the technology – its obviously sound and there isn’t an enormous learning curve. It may be a bit limiting because of Python/Django combination but again its not hard to come up to speed. Maybe its the fact that it is still in “preview mode” where Amazon’s offering looks more baked?
One very beneficial side-effect of providing a lower-level set of offerings (IaaS vs. PaaS) on EC2 is that it has created a vacuum that has been filled by companies like RightScale, SOASTA and the dozens of other companies that either lay on top of or integrate with EC2. These companies and their marketing departments are happily feeding the cloud hype monster in a way that a single company like Amazon or Google could not do credibly. Even a more comprehensive offering like Microsoft’s Azure doesn’t get the consistent coverage that Amazon is getting because of the lack of an ecosystem. Maybe being at the bottom of the stack isn’t such a bad place to be afterall.
One of the guys I regularly follow in the IT Management and Cloud Computing blogosphere is John M. Willis. John has become one of my regular passengers as I make the 45-minute drive into work each day. On a recent commute I was catching up on some of John’s CloudCafe podcasts and listened as he spoke with the Nikita Ivanovast of GridGain. GridGain is an open source platform that allows Java developers to easily write applications that can distribute their computing functionality across a bunch of nodes. The nodes can be local (on the same computer), on a remote computer, or in a cloud like EC2. The technology is very compelling and based on the demos they have on their site, it looks very well designed and implemented. Additionally, Nikita does a great job of providing examples of how to use GridGain that even guys like me can understand it. If you are a Java developer looking to do some Grid computing work, GridGain should be on your short list.
As I thought more about grid computing, clouds, and the enterprise I began to wonder how relevant grid computing will be in the very near future. I went out to Dell’s website and did a quick seach to find a high-end rack server with the most cores I could buy. In 5 minutes I spec’d out a 24-core server for around $30k. As I thought more about it, up popped an InfoWeek email with a link to an article by Antone Gonsalves called Multi-Core Processors Outpacing Key Business Software. Antone discusses the troubles businesses are having with the rapid pace at which chip makers are doubling the number of cores in their processors every 2 years and how products like VMware can’t even use all of the processors in servers shipping today.
Antone goes on to say “this year’s 32-socket, high-end server with eight-core chips in each socket would deliver 256 processors. In two years, with 16 processors per socket expected in the market, the machine swells to 512 processors total. Four years from now, the server would host 1,024 processors.”
It seems to me that if in four years we will have servers with 1,024 processors, the need for grid computing is going to slow considerably. That said, software both internally and commercially developed will need to be able to utilize all of those cores and I/O will likely become a larger bottleneck. So, will multi-core processors kill the need for most small and medium sized grid computing needs??? Should the software engineering community focus less on solutions like GridGain and more on tools or languages like Erlang that can better make use of these processing powerhouses???
So after a little over 4 years of terrestrial traveling, I finally started flying again. My instructor Gary and I went up in a Diamond DA 40 from Streamline Aviation at Austin Bergstrom (KAUS) on Tuesday. It was great to be flying again but I was a bit surprised at just how rusty I had become. The basics of flying were OK but the procedures and radio calls were pretty bad. I felt way behind the plane and looking at the myriad of options on the Garmin G1000 didn’t help the situation. It was hard to retrain my eyes to find things that previously felt so natural when flying steam gauges. The weather was clear but a front was making its way towards Austin so the winds were out of the south around 20kts with gusts reported on ATIS at 26kts.
Winds 190@20G26
The Garmin G1000
We headed east out to the practice area and started reviewing many of the usual VFR procedures like slow flight, power on/off stalls, steep turns, and unusual attitudes as I got used to the Diamond. The plane is responsive and climbs really well (especially with only two on board and strong headwinds on takeoff). After about an hour of air work we headed further east over to Giddings, Texas (KGYB) for some pattern practice. The winds were the same at Giddings so we only did a handful of landings before heading back to Austin. After this first flight I realized that it was going to take a little longer than I expected to get back to where I was at 4 years ago.
On Thursday we went flying again and the mental aspects of flying started to feel a bit more comfortable. The winds had swung around to the north at 15-20kts but it wasn’t nearly as gusty as Tuesday. Thankfully, my head was working a bit more quickly as things began to come back. We started again with airwork and then went down to Lockhart, Texas (50R) for more landing practice. Good time ahead.
Concentrating
Gary snapping pics during slow flight.
Lockhart is a great little airport that is just about 20nm south of Austin so as soon as we left the pattern and were established in the climb we began talking to the controllers. Towards the end of the flight I started to feel comfortable again with radio procedures and the G1000 was a little less daunting. We landed straight-in on 35R and as were switched over to ground we heard the controller telling a Continental MD-80 that they were going to have to wait on the ground at the end of the taxiway for 45 minutes because of a ATC hold in Houston. You just know the stress level of the passengers on that plane just went way up as they began to determine whether they would miss their connections in Houston.
Gary and I are planning to go up again around Christmas and hopefully I’ll be signed off on my BFR. After that, some more lessons to get back to where I was at when I passed my Commercial License and then onto the IPC. Somewhere in the middle of all that, Tracy, the girls and I want to start taking some trips to get Katelin and Alissa comfortable in the plane.
Back in late October the The Programmable Web quietly reached a milestone as its catalog of web service APIs crossed the 1000 mark. That may not seem like a significant number but consider that now over 60% of eBay listings come from their APIs which adds up to over 6 billion API calls per month with approximately 70,000 developers in the program and 12,000 third party applications (up from 4,800 in Q1 2007). Twitter, the free social messaging service, has an API that does 10x the traffic of their website primarily through 3rd party application integration. Facebook launched its Facebook Platform back May 2007 and now has over 52,000 applications with 140 new applications added per day, and back in October 2007 at the Web 2.0 Summit Amazon’s Adam Selipsky, VP for Product Management and Developer Relations, reported that Amazon Web Services were seeing 27,601 transactions per second. These numbers illustration that over the last couple years there has been explosive growth in the use of web API services and the pressure continues to build on software companies to open up their products to third party applications.
Some relief has come as initiatives have been launched to try to standardize these disparate API’s. Even President-elect Barack Obama is promising to “put government data online in universally accessible formats”. Google has massed its seemingly inexhaustible set of resources and placed some of them behind the OpenSocial platform to define a common API for social applications across multiple websites. A bunch of websites have already or are in the process of implementing the OpenSocial platform (e.g. Friendster, LinkedIn, MySpace, Plaxo and Salesforce.com). On the search front, Microsoft just released a re-architected API for Windows Live Search that now supports returning results in the OpenSearch format. OAuth and OpenID continue to gain popularity as standard methods for authentication and authorization and the Portable Contacts API is trying to provide a standardized way for developers to give their users a secure method to access the address books and friends lists they have built up all over the web. And finally, in an attempt to pull all of this together, the Open Web Foundation was announced as OSCON in July to create a place for community-driven specification and to support the notion of “the Web as a platform”.
It’s great to see how the technologies of the web continue to grow and mature at such an incredible rate. As more and more companies open up and standardize their APIs, our ability to deliver new and compelling services will continue to expand.
One of the most surprising leaps forward in the application of software into a new domain can be seen in the avionics market for light, general aviation aircraft. During my 15 years in software, most of the advancements I have followed are very linear evolutions of existing technology. In fact, most are re-spins of old concepts using new techniques that don’t really change how things are fundamentally done. They may change the speed or ease at accomplishing the task, but they don’t typically revolutionize things so much that they require the user to completely reorient themselves. The application of software into the cockpit of modern aircraft has fundamentally changed the mental aspects of flying.
I was actively flying until around the fall of 2004 when babies and work changed my priorities. During that time I continued to follow aviation by reading magazines, subscribing to newsletters, web surfing, and talking to everyone I could that would show the slightest bit of interest. The last plane I was flying was a loaded 2001 Beechcraft A36 Bonanza. The Bonanza is a six-seat, single-engine piston airplane that cruises at around 200 mph for about 800 miles before needing to refuel. The plane had all of the latest features that you could stuff into a single engine piston aircraft – two Garmin GPSs coupled to the autopilot, lightning detection (to help you avoid flying into a thunderstorm), traffic collision avoidance system (to help you avoid flying into another airplane), horizontal situation indicator, flight director, etc. Roll the clock forward to 2008 and the amount of information and situational awareness that is now standard on light airplanes is absolutely breathtaking.
A few weeks ago I was able to fly a 2008 Diamondstar DA 40XLS. Although a little smaller than the Bonanza, the DA 40 is technically light-years ahead of the 2001 Beechcraft. The most noticeable change inside the cockpit is that the old six-pack of “steam gauges” was replaced with a pair of 10 inch LCD screens from Garmin Ltd. The Primary Flight Display (PFD) shows altitude, speed, rate of ascent/descent, heading, turn coordination, and aircraft attitude plus a host of other secondary information such as the radio frequencies you are monitoring, distance to your next waypoint, groundspeed, wind speed/direction, etc.
What’s even more impressive is that Garmin just released a software upgrade that adds “synthetic vision” to the display. Garmin took US geological data and turned it into a three dimensional display of the topography that you see when you look outside the window (click here to see it in action – its really cool technology!). The terrain on the LCD is color coded and continuously updates to show the pilot whether you are above or below the surrounding terrain. On my recent flight we took off from Lago Vista Airport (KRYW) which sits at the top of the hills on the north side of Lake Travis. We flew out over the lake and pointed the nose of the airplane down at the water and surrounding hills side. Immediately the blue water and brown hills turned a bright red and a woman’s voice came over my headset repeating “terrain, pull-up…terrain, pull-up”. These types of visual and aural warnings can be lifesavers when flying into areas with high terrain at night or in bad weather. There have been countless airplane crashes (commercial, military and private) where pilots tuned a navigation aid to the wrong frequency and flew an approach right into the side of a mountain in darkness or while in clouds. One of my flight instructors was a Navy Captain and as his student he told me a story where a buddy of his flew an F14 into the side of a cliff on Catalina Island because he incorrectly tuned-in the wrong navigation aid. The pilot thought he was flying an approach into North Island Naval Air Station in Coronado just outside of San Diego but actually was following localizer on the approach to Avalon. The weather that night was bad and Avalon sits about 2500 feet above the ocean on top of a cliff. Coronado is at sea level so the plane descended through the marine layer and ran right into the side of a cliff. Another more publicized example was when John Kennedy Jr. became disoriented flying near Martha’s Vineyard and put his Piper into a graveyard spiral. The NTSB found no evidence of mechanical malfunction and determined that the probable cause was “the pilot’s failure to maintain control of the airplane during a descent over water at night, which was a result of spatial disorientation. Factors in the accident were haze, and the dark night.”
The PDF also displays towers (radio, microwave, cell, etc.) on the screen and color codes them depending on your relative altitude. Additionally, other aircraft are displayed on the screen right where they are at in the sky. I often complained about the collision avoidance system in the older Bonanza because I would get the audible warning of “traffic, traffic” in my headset and would be frantically searching the sky to try to find the plane. The small 3.5” display show the relative bearing to my plane with the altitude difference in feet. Probably half of the times I got the warning I never saw that actual target.
One of the new feature on the PDF was the Highway in the Sky (HITS). If you enter a flight plan into the Garmin the PDF displays a set of purple boxes in space that you simply fly through. If you point the nose of the plane through each of the boxes you will be guided right down to the runway where your flight plan ends.
Sitting to the right of the PDF is the Multi-Function Display (MDF). The MDF has “pages” that show things like aircraft engine parameters (temperatures, RPM, manifold pressure, etc.), map views, airport diagrams where the airplane’s position is superimposed on the exact location, XM Satellite Weather, approach charts, and a bunch of other stuff I didn’t get a chance to play with.
The unfortunate side-effect of all of this technology is that pilots are reporting that they are focusing so much on the information being displayed that they are not looking outside the cockpit. The amount of raw data and the slick UI can easily lure you into what appears to be a sophisticated video game. Aside from being dangerous, the reason most of us learn to fly is that we want to get away from things like computers and look at the world in the same way that the birds do.
What software market comes to mind when you think of products that:
Runs on a hybrid SaaS deployment model.
Have +10M users on a single product.
Regularly serve tens of thousands of simultaneous users.
Takes advantage of and often drive new hardware technologies.
Can be bought for a one-time fee of around $50.
Bring in $500M in sales during the first week of availability.
Have outstanding quality and usability.
Can take 3-5 years from conception to v1.0 GA.
Have a projected 2008 market of $21 – $23B
The video gaming industry has evolved into the premiere example of what software engineering could be. I don’t know how many of you are active gamers but the realism, quality, scalability and product stickiness in today’s games is incredible. I’m not a huge gamer but I am amazed that World of Warcraft surpassed 10M users in Jan (that’s 10M people who pay ~$15 per month to play online – just think about that recurring revenue stream). If you are a sports junkie and want to be successful playing Madden NFL, you need to get comfortable reading the defense, calling audibles, looking away defenders and pump-faking to get corners to bite…and that’s just for starters…when I try to play I get sacked more times than David Carr did in his 2002 season. And finally, playing Call of Duty 4 can consume you so much that you begin to get that funny feeling in your stomach because some gamer in Venezuela is hunting you down in revenge for the air strike you called in on his location. Hearing footsteps in your home theater surround speakers behind you causes you to get a bit panicky. You peek out of the second story window to get a look around and “pop”. He got you and you can hear him laughing through the voice chat system and the system replays the whole event from his perspective on the “kill cam”.
It’s insanely good software engineering. The speed of play, cinematic realism, depth of the story line, game physics, sound, artificial intelligence, multiplayer environments, etc. If you want to understand the true state-of-the-art in software, don’t look at what is coming out of Google. Those are toy apps compared to what the best houses in the gaming market are doing every day. If you haven’t played any games in years, check them out. Just be prepared to get sucked in and spend some serious time and money in the process.
I plan on using this blog as a journal of thoughts, discoveries, and ideas related to two of the things in my life that I am very passionate about – aviation and technology. Occasionally I may stray off those topics – probably on rant – but that should be more of the exception than the rule.
If you know me personally you may wonder why I did not list my wife Tracy or kids, Katelin and Alissa, as something I am passionate about. Well, Tracy and I have another family blog that we (mostly Tracy) use to keep people up to date with our activities at home.
I hope that friends, family and others enjoy the blog and find some of its content entertaining and insightful.
