Hello all,
Currently we are evaluating NoSQL for storing the time-line series data
1. Examples for storing/querying time series data.
2. Would like to know about the throughput implications especially for range queries and aggregate functions, Is it possible to stream and paginate.
3. What the table design best practices for this type of use case.
Just to give hint, here the table structure and query characteristics:
tables
- counter (source, timestamp, name, count)
- meter (source,timestamp,name,count,mean_rate,m1_rate,m5_rate,m15_rate)
Queries could be
- Get the counter stats for b/w from_range to to_range with the interval (SECONDS, MINUTES,HOUR,DAY…)
- Get count bucketed range of value – i.e. [2AM – 3AM] -30, [3AM – 4AM] =30 …
Basically want to achieve the kind of use cases that elastic search provides for aggregations.
Ex: https://www.elastic.co/guide/en/elasticsearch/client/java-api/current/java-aggs.html & https://www.elastic.co/blog/timelion-timeline
Is that feasible?
Is the assumption that Oracle NoSQL for this beats RDBMS in this case both storage and query costs correct
Thanks,
Praveen