Fork me on GitHub
#clojure-uk
<
2016-04-27
>
agile_geek06:04:59

@martintrojer: interested in your experience with Cass. I've recently carried out a proof of concept around non functional reqts with HBase and found exactly same issues. I came to same conclusion that cluster needed to be much large w/ a lot more redundancy to cope with high write alongside reads. PoC finished before I had chance to test theory though!

agile_geek06:04:45

Getting an even split of data across region servers also made a huge difference but it was very delicate balance and different data profiles made huge difference

agile_geek07:04:48

Getting an even split of data across region servers also made a huge difference but it was very delicate balance and different data profiles made huge difference

martintrojer07:04:54

Yeah, and why put you though all that pain when you can just use Dynamo?

martintrojer07:04:19

Its like RDS just removed all the pain of setting up PSQL read/write slaves etc

otfrom08:04:22

martintrojer: I suppose the difference between RDS and Dynamo is that Postgres is Postgres (and you can get it other places). Dynamo is not Cassandra (or anything else). Not a reason against it, but a trade off to think about.

martintrojer08:04:44

its a cracking service none the less

otfrom08:04:31

of that I have no doubt

otfrom08:04:54

that's the problem with tech. The answer is almost always "it depends..."

yogidevbear10:04:19

G'day everyone 👋