Answer the question
In order to leave comments, you need to log in
Best DB for storing graphs (values)?
A huge amount of data has been accumulated. Data from devices in the form of "position-indication".
More than 10,000,000 such graphs have already been accumulated, each of which contains an average of 3200 "position-indication" pairs.
Approximately 2 million graphs a year are growing.
At the moment, everything is stored in such a structure in the Oracle relational database: a description of the graph in a table in one table, in another table a blob is stored with a text file in an archived form where values and an identifier from the first table are stored.
This structure does not satisfy in many respects. Firstly, comparing graphs with each other is impossible without an additional tool (currently Java). Secondly, it takes a long time, you have to extract huge arrays from the blob, then pump them out to the server where the Java machine is running and process them there. Secondly, it is inconvenient to save them to disk as files for transfer to users.
We decided to try to store all the values in a regular relational structure, but we extracted 5% and realized that there would not be enough disk space, and I would assume that the speed of work would leave much to be desired.
The database should not be a temporary storage, but quite an archive, from which, on a daily basis, users will receive information in the form of files to their PC.
Are there any NoSQL databases suitable for this task?
Answer the question
In order to leave comments, you need to log in
https://en.wikipedia.org/wiki/Time_series_database - database class for your tasks. This is of course if you have time to try and you are already convinced that the relational one will not suit you.
Didn't find what you were looking for?
Ask your questionAsk a Question
731 491 924 answers to any question