I don’t like most opensource time-series databases that have been created recently. Some of them lack compression, some of them are slow or focus on the wrong problems. In my opinion good time-series database should satisfy several requirements:
You can argue that a system like this is not very useful and a complex query language or other features should be implemented. Well, this is a very basic requirements to start from. I’m planning one advanced feature that absent from all time-series databases that I know — indexing. Not the same indexing as used in normal databases but specific for numeric time-series that can speed up similarity search on numeric time-series data. Such indexing opens paths for things like time-series clustering, motif discovery, summarization and anomaly detection. I think that having the ability to quickly perform similarity search is infinitely better than the ability to execute continuous queries or draw graphs.