It's been posted in the comments below, but I'll reiterate it because it needs to be said. This article is more about the issues of naive database adoption strategies than any deep fundamental flaw with Cassandra.
I've deployed C* on some similarly large datasets and encountered none of these issues - even when storing terabytes of time-series data. The difference - I read through the documentation from head to foot before getting started, did numerous dry-runs to figure out the data modeling, and when I wasn't sure about something asked the community (who are usually incredibly quick to respond on IRC, twitter, or elsewhere).
Comments
It's been posted in the comments below, but I'll reiterate it because it needs to be said. This article is more about the issues of naive database adoption strategies than any deep fundamental flaw with Cassandra.
I've deployed C* on some similarly large datasets and encountered none of these issues - even when storing terabytes of time-series data. The difference - I read through the documentation from head to foot before getting started, did numerous dry-runs to figure out the data modeling, and when I wasn't sure about something asked the community (who are usually incredibly quick to respond on IRC, twitter, or elsewhere).