The Amazon Technology Platform
Earlier this year my good friend and mentor Jim Gray dropped by at Amazon to interview me for ACM Queue magazine. I had real fun with Jim’s questions and I like to believe that the resulting interview is interesting. It appeared in this month’s edition and runs about 8 pages in the print edition.
It interview touches on many different topics around the Amazon Platform and I believe that there is quite a bit of information in the article that has never been publicized before. We get into Amazon’s technology history, scalability of services, the service architecture, large scale testing, large scale development, the retail and enterprise partner platforms, hiring, the relationship with academia, and many more.
3 TrackBacks
Listed below are links to blogs that reference this entry: The Amazon Technology Platform.
TrackBack URL for this entry: http://mt.vogels.net/mt-tb.cgi/29
ACM Queue has just released an interview of Amazon’s CTO Werner Vogels. He is an ancient from Cornell University and is specialised on HA and scalable system (see his thesis work for more information) I read his blog regularly since he addresses ... Read More
The talented Werner Vogels’ of amazon.com has just been interviewed. A quick scan revealed a very interesting article. The full article is on my reading list. Once the EBA AJAX Grid component is released the article will make a great read. ... Read More
Track: Web/Services Code: WSV203 What I call the config dilemma of Enterprise Services is Enterprise Information Technolo... Read More
Subscribe to this weblog's
This is really an interesting read.
It does make me think about sending my resume to Amazon :p
Interesting interview, Werner. Can I ask a question?
At one point, you said, "A service-oriented architecture would give us the level of isolation that would allow us to build many software components rapidly and independently."
At another point, you said, "If you have this decentralized organization where everybody is developing things in parallel, how can you make sure that all the pieces work together as intended ... Testing in a very large-scale distributed setting is a major challenge."
These two statements seem to be in conflict. Are the services independent of each other? Or are there dependencies that complicate testing?
This question goes to a broader point. You described the problem with the architecture in 2001 as being a complicated nest of dependencies with poor isolation. You present a web services architecture as the solution.
But why do web services necessarily eliminate dependencies? Can you not have the same complicated nest of dependencies in a cloud of web services, all of them talking to each other in an incomprehensible chatter?
Your unnamed reference to some of the missing assumptions in existing work on decentralized systems is teasing.
I'm curious if Amazon have been looking at decentralised routing techniques to route messages internally to help avoid scalability issues associated with using centralised naming services.
Is this not just a form of decentralised Enterprise Service Bus, not unlike Artix?