0

I have a time-series dataset (around 500 million rows, 4 columns) that stores public transport ticket validations.

Not sure if it matters, but it's an event driven dataset, without constant timestamp intervals, and with possibly overlapping timestamps.

The data will be written only once, so write speed is not important to me.

My work is mostly related to data analysis and processing, so a high read speed is crucial.

Having almost 0 database experience, I started looking into popular time-series databases and experimented with influxdb. However, reads were excruciatingly slow.

From what I understand, influxdb (and other time-series databases) is designed to support high write speeds, which I don't need.

What database fits my requirements?

EDIT:

3 columns are ints (4 bytes), 1 column is a bool

Improve this question
3
  • Please give us the structure of this dataset - i.e. the datatype (with size) of each column. FWIW, I think that the best all-round database for a novice is PostgreSQL and unless you're having problems with it, then keep using it. I would also recommend that you benchmark a couple (perhaps 3) solutions and go from there. p.s. welcome to the forum! :-)
    – Vérace
    Feb 24, 2020 at 10:25
  • @Vérace thank you, I've added the structure.
    – ruiloureiro
    Feb 24, 2020 at 10:34
  • So, approx. 7.5GB of data - any RDBMS could cope with this easily - 7.5GB is "small" (a relative term of course...). I'd recommend PostgreSQL (with an SSD) as my first port of call - if you're having specific problems with it - ask a new question here.
    – Vérace
    Feb 24, 2020 at 10:38

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.