Grokking the System Design Interview
Vote

0% completed

Read Heavy vs Write Heavy System

Two systems can store the same kind of data and still need opposite designs. A news site serves a hundred reads for every write. A metrics pipeline takes in a thousand writes for every read.

So before you choose a database or a caching plan, answer one question. Is the traffic mostly reads, or mostly writes?

A read-heavy system serves far more reads than writes. News sites, product catalogs, social feeds, most public web applications. A write-heavy system takes in far more writes than it serves reads. Logging and metrics pipelines, sensor ingestion, click tracking, chat delivery.

.....

.....

.....

Like the course? Get enrolled and start learning!
A

anmoldeep1509

· 4 months ago

Write-Ahead Logging (WAL) is given as a strategy to improve write performance, but I think it's the opposite. Because in write-ahead logging, we first write to the cache and then to the database. Hence, writing to cache is synchronous and adds time to the write operations.

Instead WAL is a strategy that improves the read performance, because in this case read operations are from the cache (which is much faster then database read).

Show 1 reply
B

basaranbahadir

· 2 years ago

Optimize database schema and indexes to improve write performance.

stated in Section Designing for Write-Heavy System.

I guess you explained the opposite in the previous course, by saying that creating an index is beneficial for read-heavy databases, not for write-heavy ones because when any update is requested on data, the index should be updated as well, hence it increases the latency.

Show 1 reply

Reading Progress

0%