The team from Amazon Web Services (AWS) came to Toronto (at the MaRS Building)to show us their stuff, and demo’s from some local folks showing the “what”, “where”, and “how” to their use of AWS in real life. (no admission cost for the event. Yea!)
Tracy Laxdal did a great job on all the the logistics and organizing.
Prashant Sridharan and Mike Culver, the Director and Evangelist for Amazon Web Services (respectively) kicked off things with comments and a presentation giving some background to the history of AWS, where things are too today and where you can expect this to go.
The overall focus was on the Infrastructure Services : in particular things like S3 Storage and Elastic Compute Cloud (EC2), but it did touch on the other parts (which is good because it did clarify some of them for me).
Update Nov 18th : Amazon’s CDN is out, via GigaOm : Amazon’s CloudFront Could Storm Rival CDNs, with more details on the AWS blog” Distribute Your Content With Amazon CloudFront and Amazon CTO Werner Vogel’s blog Expanding the Cloud: Amazon CloudFront. Go Stormly with cloudfront.
The meat of the afternoon was in Customer Presentations :
Carlos Barrettara is from Polar Mobile, a solution for publishers and adversiters on mobile platforms which -for example -would allow BlackBerry users to receive the latest headlines, news articles, feature columns, and editorials directly from Maclean’s and Canadian Business. As a very new startup they have been able to embrace AWS from the start.
Carlos’s presentation is now on SlideShare.
Ilya Grigorik is from AideRSS, which is a site and add-on and web service which acts as an intelligent RSS assistant to allow you to focus on the most information content of the Websites and Blogs you follow in your News Readers.
Ilya talked about how they moved (in great detail, with numbers of servers : 14 for infrastructure and 70 (!) for web crawlers) the various parts onto the AWS cloud (including using Simple Queue Service ) as they gained comfort with it. Ilya’s presentation is now on Slide Share. (also their presentation from last year)
He also showed off the custom scripting he uses. We should get Chris out to a Demo Camp! Chris’s presentation is now on Slide Share.
Farhan Thawar is from I Love Rewards whom enable corporate incentive programs.
They moved from a traditional self hosting infrastructure to the amazon cloud for all but some legacy application. Given they are a Software as a Service and integrate with SalesForce (another SaaS) it made sense to embrace this style of hosting. Farhan’s presenation is now on Slide Share.
Paul Bloore from Idee talked about their newest product TinEye, (I promise I will stop call it TinyEye some day, Paul!) their image search engine that tells you where and how that image appears all over the web—even if it has been modified.
Paul talked about how they used EC2 to build a spider army to scan the web for new images, (and process the imaging matching secret sauce?) (currently at 900 million, soon to be 1Giga+! images indexed, he said in this best Doctor Evil voice.) Then he showed of a unreleased iphone app that takes a book or cd cover and uses a similar algorithm as TinEye to identify it and fetches reviews. Which caused my brain melted down ….
(Update : see TinEye Mobile : iPhone visual search for stuff )
The Start-up Tour is well worth your time, and is still to come to New York City, Boston/Cambridge, London, Amsterdam and Seattle.
Some of the Presentation from last year went up on SlideShare (which uses AWS), so I hope this years will too!
34 on the Toronto presentations are now on SlideShare the startup project, plus those for other locations.
Now excuse me will I do some Cloud Hacking…
minor updates :
other posts about the amazon web services start up tour in Toronto :
- TechVibes on Amazon Web Services Startup Event in Toronto
- Luke Rodgers aka everything flows on Amazon Web Services Start-Up Tour in Toronto
- Foobartastic on Amazon’s Cloud looks to solve many problems
and thinking about “use cases which are different form the usual uses” :
- batch processing, either semi temporary or semi permanent (how often do temporary – or short temp – system and applications last for years and years? often), when you need 1 or more (or 100) computers to process a set of data needed by other processes.
- those semi temporary or semi permanent applications that are needed until the real solution comes along
- Transitioning from one version of the infrastructure to the next, one customer at a time, and still roll back to the previous version.
- Real testing infrastructure that actually look like the production infrastructure.
- thinking about some side discussions I had with Chris Nolan who has a Comic Books application on the FaceBook platform (see the about for more info), I though of anther scenario : if you know you have big spikes in the traffic or processing demands : one day of the week -new comics come out on Tuesday- or year -election day-, or Month – Xmas-, then add servers only for that known period (and only pay for them during on that period). You could also monitor loads and start adding servers auto-magically when you exceed a threshold and remove them when you go before certian thresholds.
Any other suggestions come to mind?