Skip to main content
3 votes

MySQL solution for many to many associative table that scale over billion entries

Preface I think you're unfortunately misunderstanding: How databases work at their core level, and What some of the people of this community have been trying to communicate to you in regards to that. ...
J.D.'s user avatar
  • 38.1k
2 votes

Postgres jsonb vs composite type performance differences

Edit: These tests were performed in Postgres 12.16. When I ran the queries in the selected answer against my own data, I found significantly different results. In particular, the LATERAL JOIN to ...
Robert's user avatar
  • 123
2 votes

MySQL solution for many to many associative table that scale over billion entries

The best solution for your problem is what you already designed. That's because all RDBMS are designed to handle exactly that: a bunch of tables, with many-to-many relations, and potentially billions ...
Andrea B.'s user avatar
  • 1,613
1 vote

MySQL solution for many to many associative table that scale over billion entries

Since there are only 100 items... If there won't be a lot more items, change to SMALL UNSIGNED (max of 65K). This will save 30GB. (Shrinking a table --> less I/O and better caching --> maybe ...
Rick James's user avatar
  • 78.6k
1 vote
Accepted

Setting up Transactional Replication - Can i safely remove transactional replication if it is causing issues on my publisher?

Can i safely remove transactional replication if it is causing issues on my publisher? I am answering your broad question first. The answer is yes. You can follow Manually remove replication in SQL ...
SqlWorldWide's user avatar
  • 13.5k
1 vote
Accepted

Designing Tables for Large Amounts of Time-Series Data

It depends on what you plan to do with the data. 20 measurements is not a lot of columns to be honest. 1 reading per minute is not that much, and billions of rows in a table isn't something to be ...
J.D.'s user avatar
  • 38.1k
1 vote

Designing Tables for Large Amounts of Time-Series Data

I will use MQTT terms: "topic" represents the unique identifier of the sensor (for example src_id, measurement_id), "timestamp" obviously, and "value" is the measurement ...
bobflux's user avatar
  • 1,551
1 vote

Designing Tables for Large Amounts of Time-Series Data

I recommend option 1. The problem with option 2 is that each table row in PostgreSQL has an overhead of at least 23 bytes apart from the data, so the tables will become much bigger. Option 3 is ...
Laurenz Albe's user avatar
  • 53.5k

Only top scored, non community-wiki answers of a minimum length are eligible