Talk comments

Anonymous at 21:26 on 15 May 2011

I had the answers to any my questions. Thanks.

This talk gives an overview over the various concepts that are summed up under the term NoSQL Databases, their concepts in an almost mathematical detail, their pros and cons, and some (open-source) implementation of each kind. Possibly also due to the unrewarding time slot at the end of the conference, it seemed like with this expansive approach the talk wanted to achieve too much, and the slides will be more helpful as a reference than to actually support the understanding of the subject matter. Most important take-away for me: the term NoSQL really mostly just says what it *not* is, rather than being really useful in its own.

A nice presentation of how you can use CouchDB to build performant and fail-safe applications and how its reliable multi-master synchronisation allows you to move data between several applications in separate locations (including iPhones). And how its JSON interface allows you to build front-end applicatoins right on top of it. Would have been interesting to learn more about practical aspects like the authentication stuff that was mentioned towards the end. All in all a good and encouraging start to get familiar with CouchDB for me.

This talk provides a good overview over many effective measures to accelerate a slow web site. Most of them are not really new as suggested by the title, though, and a few ideas could do with some refinement. It may mostly serve as a good starting point and introduction.

Everybody knows what REST is all about, and we know what Hypermedia is. Or do we? This talk is an eye-opener. You might see right away how you got API versioning wrong, or content format negotiation. Most importantly, however, you'll see how REST means more than just having four operations and a resource and you might realise, as I did, that your grasp of what's really behind this powerful concept is much more limited than you might think. At least you're in "good" company with Twitter and many others--if that doesn't ease you, see this talk and go learn what the Uniform Interface is all about. Thanks, David!

A concise introduction to the concept and implementation of Varnish as a standards-compliant and very effective reverse caching proxy (or http accelerator, as they like to put it). Thijs' doing a short part about getting it to run and emphasis on how to subsequently (and continuously) understand what Varnish does and optimise it for its individual job to me perfectly matches the way how such a tool should be operated. This all was presented in a very understandable and entertaining manner. An excellent talk!

DI means passing objects to other objects while we build them instead of instanciating one inside the other. This talk nicely shows some of the various concepts of how a developer can define which object, or which type of object, is to be injected where. What I did not get from it is how these approaches relate to each other. It seems like there's some evolutuionary development of which they represent different stages, but still each of them will be the right one to use in certain situations. But there was definiteily some food for thought in this talk.

A very useful introduction into the reasoning behind Sf2's concept (and good implementation) of leaving the caching to the http layer and protocol rendering it both more effective and more maintainable (which also means more scaleable) than non-standardised alternatives. The ESI part was a bit short though, given the talk's title.