ITA Emerges from Google's Shadow, Unveils New Airline Platform | Hacker News Reader

ITA Emerges from Google's Shadow, Unveils New Airline Platform

xconomy.com

52 pointsgthuang130 comments

I'm particularly interested to see how the airlines react to the re-emergence of ITA and its new airline platform. For years, airlines have essentially had one choice in reservation systems: Sabre. Sure, there are other systems, but none of them offer the functional breadth that Sabre provides -- including ITA, of course. But where it may get interesting is that, according to a colleague who works at Sabre, many of their customers are growing increasingly frustrated with large numbers of unaddressed bugs, lagging system response time, and overall system instability. Some airlines may decide they've had enough and jump to ITA, and even if ITA isn't successful at wooing away some of Sabre's customers, hopefully the new platform will at least spur Sabre enough to improve its aging product offering.

Do Amadeus and Travelport not also host airlines? And there is HP/EDS who also host some major airlines. Plus SITA, which hosts many smaller airlines.

Regardless, all of these systems are very ancient. It's refreshing to see a new generation emerge.

Related: I really like ITA's flight search interface. It does everything I need, correctly.

http://matrix.itasoftware.com/

Then you're gonna love Google Flights:

http://flights.google.com

The obvious thing google flights misses over ITA is the month-long search option, which I use a lot. For the power users, ITA has a very useful regex-like syntax for specifiying flight queries, but that is understandably not of as much interest to the vast majority of people

I live and die by the "advanced routing codes" on Matrix... see my previous post here:

     SFO,LAX CX,AA,KE F HKG,ICN,NRT ~CZ+ KMG

http://news.ycombinator.com/item?id=3090362

Google Flight Search has a calendar view: it's hidden behind the button directly to the right of the return date and lets you play with different departure date and trip lengths.

"Sorry, locations outside the U.S. are currently not supported."

Well, I'm sure it's a very nice interface.

Yeah, I just tried to search a flight to a European country and it said "unavailable". So it's not accessible from the US to the outside either.

It takes time to support all the airlines in the world. Give it a few months...

No, it doesn't. Every major airline publishes their fares, schedules, and availability in a standardized format; furthermore, all of this data is available from a single collection point. If Google wanted to allow international searches, they (or more importantly, ITA) already have everything they need.

They quite literally just need to "flip a switch" to enable it.

Most likely the reason they don't is that each additional origin or destination city that they add requires additional servers and storage and computation, and so it's much more practical for them to limit the tool to the top 50 or so Domestic USA airports.

Not so much.

If you've used flight search, you'll have seen that it's instant response. Change a filter on price, length of flight... anything... and the results instantly change.

The results are pre-computed... thus extending it to international flights is far from just "flipping a switch". It makes the pre-computation 1-2 orders of magnitude more difficult!

Thank you for this info very useful. Where/what is this "single collection point"?

The schedules come from OAG, the fares come from ATPCO, and the availability (number of seats available on each flight in each booking class) comes either directly from the airline or through one of the three GDS providers: Sabre, Amadeus, and Travelport.

Technically, the schedules and fares are also available from the GDS, but only in an interactive format. ATPCO and OAG supply giant dumps of bulk data which is more suitable for offline processing and precomputing results.

ITA backend supports them already (e.g. see the matrix site), it's just not as fast as Google flight search.

I've been a confirmed Kayak fan.

I don't need to go anywhere at the moment, but you comment got me to try a search on Google Flights. Let me just say, Wow! Simplicity, organization, speed!

Of course, the true test will be when I do have to book. I'll be sure to try 'em both out.

Partially to spur conversation and partially because I don't remember the details of previous discussions, does ITA still use LISP?

I interviewed there recently, though decided to take another job. ITA uses Lisp for their core flight search engine, but uses other languages like Java and C++ for many of their other projects, such as the one mentioned in this announcement.

ITA developed the airline reservation's business logic in Common Lisp. They were using Clozure Common Lisp for that.

When I interviewed there in early 2010 they still used LISP. They said it was mostly because their hires fresh out of MIT wanted to use it, so I don't know whether they will continue to use it now that their MIT kids will no longer want to use it...

Scheme is no longer used in the introductory EECS course, but that doesn't mean there won't still be a lot of MIT grads hooked on the Lisp family of languages. (I think ITA uses Common Lisp.)

I imagine rewriting a large code base in any language to be very expensive, I doubt somebody can justify such an expense just because a bunch of new hires don't want to use a language. Companies don't rewrite java apps in lisp because a new of college grads don't want to use java.

No matter how much I wish they would...

I can't say i disagree with BigCo on this one. If one wishes to use better languages the options are to switch jobs, start your own company or stay, but push to augment the existing codebase with other JVM languages. Clojure and Scala people have had some success starting skunkworks projects. But a complete switch? In most cases thats insanity. Companies have actually died trying to rewrite their apps from scratch. And even a slow gradual rewrite might take too long and be too expensive, not to mention not bringing any immediate benefits, because you're not adding new functionality and fixing existing bugs.

From their homepage: http://itasoftware.com/careers/work-at-ita/ita-engineering.h...

"We code in whatever language is best suited to the task, be it Lisp, C++, Java, Python, Perl, or XSLT."

They still list LISP in the job description for ITA.

I work at Google on flight search and was at ITA before that. We indeed continue to use Common Lisp. QPX, the flight search engine, uses SBCL.

I think Google should just experiment and dabble in every single niche in the internet field.. just push the envelope. Start a dating site to compete with Match. Start an auction site to compete with eBay. Start a ticket exchange to compete with StubHub. Just throw a bunch of stuff in the wall, and I'm sure a new billion dollar business will emerge.

This was the strategy that they with Schmidt. Page changed it and shut down a lot of products to focus on the company core. The shotgun approach seldom works in business. It is better to develop and nurture cash cows.

Personally, I think that they should focus in the search field to stay ahead of the curve and prevent a competitor from introducing an iPhone caliber product that would make them look myopic like Blackberry. They used to claim that search was a 99% solved problem. I don't think that me looking to a list of blue links is 99% solved problem.

Amadeus employee here..

Indeed one of the largest issues we face is not just building a new reservations, inventory, departure control platform on Linux but having them work in conjunction with our legacy TPF systems. Newer airline systems don't have this problem and have complete degrees of freedom. Interesting to see a new reservations platform and not just flight search coming out into the wild.

Reply on news.ycombinator.com