Which modern database to choose for BigData?

Question

MongoDB

Hello colleagues!
What database would you recommend for big data of 100 million rows or more. I want the impossible: easy to design, quick to choose, fault tolerance and scalability by means of the database itself.
At the moment I don’t know what to choose:
1) MongoDb is quite simple, there is work experience.
2) Cassandra is fast MongoDb, but there are many nuances in work, not very flexible in terms of data extraction, you need to dance with a tambourine.
Maybe there is something else?

Answer 1

You don't have BigData. Take a regular database and shard the data into several tables (this is the maximum you need).

Answer 2

First you need to decide what kind of data you have there. And why, for example, do you use InnoDB, and not MyISAM? And try MariaDB, there are other engines out of the box, maybe that will help.
As for 100GB - what do you want to keep the entire database in memory? what for ? it's enough to keep indexes + head-on advice: move data to SSD (this is if you have a lot of random reads on the database). At worst, if you need text search, then take Sphinx (its indexes take up little space).

Answer 3

Thanks for the tips, begemot_sun and portfelio!
For 100 million rows, then you will need unmeasured operatives -> a very expensive server, it’s easier to buy microinstances and do sharding and not through the application, as with Mysql.
Approximately how much memory is needed for Mysql databases (InnoDb) with a capacity of approximately 100 million rows? Depends on the size of the database itself and what would it fit in memory?
For example, now we have 11 million (Doubles every month), tables together with indexes weigh 10 GB. Recently, with 4GB of RAM, we switched to 8GB, the flight is normal, but probably in a month you will need to switch again (((.

Answer 4

begemot_sun,
We use InnoDb because it has a row-level lock, which allows you to simultaneously read and write data to the table.
InnoDb allocates memory for the data and index cache in RAM.
Thank you MariaDB, I'll take a look.