When One Machine Is Not Enough
Cutting the Data Into Pieces
Last timePutting Events In Order Without a Clock
When the data no longer fits on one machine it has to be cut up, and almost every difficulty that follows comes from the choice of where to cut.
A dataset grows past the largest disk you can buy, or a write rate grows past
what one machine can accept. Copies do not help with either, because every copy
holds everything. The data has to be cut up.
What splitting is for
It is worth separating the two arrangements clearly, because they are often
discussed together and they solve unrelated problems.
| survives losing a machin | serves more reads | holds more data | accepts more writes | |
|---|---|---|---|---|
| copies only | 0.95 | 0.90 | 0.20 | 0.25 |
| pieces only | 0.25 | 0.30 | 0.95 | 0.90 |
| both together | 0.90 | 0.85 | 0.90 | 0.85 |
The key decides everything
Every record is assigned to a piece by its key. That single choice sets how
evenly the load spreads, which records end up near each other, and which queries
stay cheap.
The lesson stops here
6 more paragraphs to go
You have read the opening. The rest of the argument, the problems that check whether it landed, and the lines worth keeping at the end all come with a plan.
The first lesson of every course in the library reads the whole way through, free, so you can see exactly what the rest of them are.
See the planThe contentsThis is the reading half
Starting the course gives you your own copy of it. Every idea on every page has problems standing under it, marked with a reason rather than a tick, and any sentence you do not believe can be opened and argued with. None of that can happen on a page nobody owns.
The contents