the subject was interesting, but very basic talk, and some information not true (e.g. Solr is not the only one based on Lucene, Elastic is another one).
It'd be better a comparison using the same metrics
- velocity of indexing (with real results)
- search velocity, including velocity during indexing
- additional features (facet etc..)
- documentation and avg time taken to learn
or the result was "this one is fast, the other is not documented, etc... not helping in selecting one
after the talk I had info that I had already
Good talk. Unfortunately run out of time when it got really interesting
I think Dave said at the beginning of the talk he was there to promote the product (or something like that) but still...
I wanted to walk out in the middle of the talk, I'm sorry... :-/
I was under the impression that this was going to be a talking regarding the pros and cons of NoSQL as a database solution vs SQL. For example, where might a document oriented database better suit a solution rather than a schema based one. Instead it was more focused on software, CouchDB vs MySQL which does help the arguments but the software can change.
Agreed that not much was learnt from this (so what was the point?) and it came across as more of an ego contest.
I have to admit, this was the talk I was most looking forward to. I expected there to be a lot of real clever things going on behind a name as big as the BBC. Instead it all seemed a little 'hacky' (but I guess also smart) solutions to cache as much as possible. This topic was the focus of the talk. Touched on dealing with errors/bugs but went no further than checking Apache error logs?! Monitoring (the topic of the talk!) didn't happen until later and even then it was rushed and unclear.
One thing that was irritating was that at the end of the talk, I had some questions and after waiting 15-20 minutes at the front, the speaker walked off towards lunch even though he saw me waiting, talked casually about something unrelated to someone else on the way and when I finally managed to talk to him (while he queued for lunch) then seemed more interested in what they were serving then communicating with me. I thought this sort of interaction was supposed to be what these conferences are about, otherwise what can I get from it that I can't get from a blog post? Very rude and disappointing. The only thing I really took away from this is maybe giving Varnish another shot as a caching tool and working more with VCL.
Load Balancers, Queues etc are not a new subject and although for someone new to the idea, they are a fantastic solution, there wasn't any important detail. For example, queues are all well and good but what happens if part of a job fails and how do you keep track of tasks? Futhermore, when discussing server elasticity and scaling, Helgi was unable to provide any ways of implementing this yourself and was unaware of the many services that can help.
In addition, the idea of suggesting "Internal Storage API" construction for easy storage switching seems a bit over the top when this can be achieved by implementing storage interface classes (and dependancy injection of course) for easy swapability of storage methods (that is what they are for!).
Very confident speaker and clear slides, but perhaps too much of a plug for what his company is doing (and partnering companies) and not enough research into the topics.