I’m fearly new to influxdb, and I’m pondering over it’s use for storage of financial data.
Like ticker data from crypto pairs.
The storage of data is quite easely to do, atm I dump ticker data using a python script over udp into an influx 1.8.
The reason I use the 1.8 version is, 1.8 supports subscriptions while 2.0 does not. (correct me if I’m wrong).
I’m searching a way to do calculations of technical indicators and/or do outlier detection on the tick data whenever a new point is available on the latest xyz ticks and push it back in a new database.
From the docs it seems kapacitor should be a possible way to achieve this, however go is not my style and the learning curve of kapacitor and TICKscript seem quite steep.
I’m struggling in v1.8 with some order by queries which seem to be only possible by time in v1.8.
I’ve noticed telegraf has streaming capabilities too, but it seems the least possible interval is 1s. Is this correct or do I miss something ? Further question does telegraf dumps points directly to the db or keeps it internal and flushes them only when a thresshold has be made (like in the telegraf docs 1s or for example 1000 points)
Atm, I’m wondering what the best approach is : staying with v1.8 and doing the group by orders external or upgrading to v2. Does v2 has something like subscriptions so I can do the ta calculations in realtime on the fly external and push them back in another field of a bucket … or … do I stay with v1.8
What could be the best approach to this dilemma ?
Second question :
Creating OHLC data from ticker data can be done by continious queries or tasks (depending the version). Continious queries do take the whole timeseries in the from database. To speed up calculation is there a way to do continious queries or tasks only on the points which where new in contrast to the previous query, or do them on specific times (like on every minute) instead of using a delta time like each minute ?