Jeff Barr spoke about Amazon Web Services yesterday!

And I was there to listen to him! Although I don't have any picture as I forgot the camera and I didn't take any note as I didn't bring a ballpen (or a laptop, as the modern speech-attendants do), I'll try to summarise here the best points of the talk.

The event was held at Westminster University, on the New Cavendish St Campus. It was the first time I visited an english university so I was quite curious about how the ambience was going to be, the installations, etc. It looked all quite modern and a bit posh, but without being overhelming. Our place was the big Lecture Theatre, and it actually resembled a lot all of those american classrooms which I only had seen in the movies. It had that feeling... Jeff was actually a very nice person, quite accessible and open to any questions and feedback. He promised to transmit all of the feedback to the working team at Amazon, as seems like these speeches are used as a source of feedback from active developers. Something like a pool of ideas for everybody. So the event was more of an open discussion between developers than a university lecture. This wasn't an inconvenient at the beginning, since the seats seemed quite comfy even if they were the austere style of classrooms seatings, but at the end after almost two hours being there my back really hurt and I was willing the people to stop asking questions and allowing me to go home! (This sensation is specially more intense if you just have been sitting all day long in work).

So now to the speech itself... he gave a good overview of how he finished working in what he works and he was quite honest "I really didn't believe I was the best person for my job". After that he introduced Amazon briefly and later, each one of the technologies which Amazon provides to the developers, either paying for the use or not. Obviously that was the longest part of the speech. I loved this sentence: "We want to open up the creativity of the developers". That is quite nice! So, the system roughly works like this: we offer you some information in a way that we both can benefit.

Allow me to remind you that I didn't take any note and my memory is not really reliable so maybe some of the names aren't exactly ok. If in doubt you'd better go and get some info by yourself!

These are the main services that Amazon offers (and I remember):

  1. ECS (E-commerce Services). That allows you to retrieve data from Amazon store products. And when they say "data" they are referring to lots of data! Even images. You can then do whatever your creativity allows you in what regards to showing that information. For example as you can track the price variations in the products, there's people which has built like an Stock Market with Amazon product prices. Other people has built an application in which you add something yo tour wishlist and when it reaches a given price (or less) the system sends you an email.
  2. Alexa services. You can access Alexa data. This is what is done for example in webs like alexaholic. You can evey buy your own search engine using Alexa services. He explained how this is done, in a way that your code is converted into parallelised code and sent to the search engines, and then you're invoiced depending on how many resources you used with your search engines. This is interesting since it allows you to have a certain functionalities but without having to invest in all the infraestructure and maintenance needed. Terms like Return of investment arise quickly here...!
  3. Mechanical Turk. I hadn't listened about that before. It's like a way to use Distributed Human Intelligence. You create "Human Intelligence Tests" (HITs) and put a price and a maximum number of replies. Then people around the world which participate in the project decide to answer your question, being payed the amount you specify. Although this looked stupid at the beginning if you think later is a very powerful medium to get massive results without having to invest in infraestructures. He put an example with image recognition: create a HIT for classifying pictures distinguishing between them having a human face on them or not.
  4. S3. I'm pretty sure everybody has listened about this storage service that is provided by Amazon. He did show us some examples built upon it, even applications which run entirely on the client side and have only persistence in S3 servers (an impressive javascript wiki).
Of all of them I would say the one which impressed me more is the Mechanical Turk! I still can't think of an immediate application for my ideas but it left me mumbling around it, which is very cool.

Also somebody asked about accessing imdb data (which Amazon owns too) but Jeff said that although it's been asked more than once, they don't have plans to allow access to that yet. If you want to find more about all of these services you just need to have a look at Amazon Web Services. There's also the Amazon Web Services blog, which is written by Jeff, where you can find lots of examples to sites which use them, and finally Jeff's blog which obviously is more personal.

I would like to thank Dean Wilson which organized everything. It was a very nice evening :)

UPDATE 23 may 2006: Steven Goodwin provides some pictures of the event. Thanks Steven!