I’m new to the influxDB and I want to use it because I heard that its suitable for handling the time-series data.
So I import a sample trading data with 80000 points into the influxdb,also build a table in mysql to cmpare the performance.
I firstly test “select * from measurement where time < Atime and time > Btime”,it shows that infludb is about 3 time slower than the mysql.
Then I test “select count（*） from measurement where time < Atime and time > Btime” and some other aggregate function，influxdb shows a dramatic improve comparing to mysql.
I want to know if this is normal,and how can I improve the query efficiency of raw data(points).
When considering which database to use for TSDB, you should look beyond trivial benchmarks like that; they can often be rather misleading. You should try the same benchmark again with a few hundred million records.
InfluxDB is purpose built for time series data, handling nanosecond resolution, sharded on time, retention policies, downsampling, Flux query language, and many more benefits.
If you wish to discuss your specific use case, I’d be happy to provide advice.
Best of luck
Thanks for the reply. But I don’t have that huge amount of data sample to handle. So do you have any benchmarks with hundred millions of records which I can refer to?
I found this https://cs.ulb.ac.be/public/_media/teaching/influxdb_2017.pdf on line which compare the performance bwtween influxdb and SQLServer on 30,000,000 records. On certain senario influxdb is just slightly fast than SqlServer. So is there any best practice on how to query the data to acheive the best performance?
