Hacker News Reader

Comment by marktangotango | Hacker News Reader

I suggest you run some simple load tests that do inserts, you'll find WAL is not sufficient for concurrency. Multiple reader, single writers handle this well though.

View on HN

You can put all your writes in a queue to be processed by a finite set of writers. I think most CRUD applications can afford the delay and if not you can pair it with a write through cache.

It sounds simpler to setup postgres and never have to set up a queue (unless you’re using an in memory queue, and that has its drawbacks).

Although I definitely would like to use SQLite just for the cost savings for something. Litefs/litestream looks great.

I was surprised at the amount of performance SQLite provided. I get about 700 inserts/sec with the WAL enabled sitting behind a Django app.

https://imgur.com/a/a3U41Zo

Without the WAL enabled, I get a around 400req/sec.

edit: clarity

Concurrent writes are the bane of SQLite's existence, but for a one-person blog, you shouldn't run into any issues.

You would have trouble scaling up to fives of authors, though, which would be a deal breaker for any serious production app.

@levelsio famously runs a dozen websites all backed by SQLite, with 200M reqs / month: https://twitter.com/levelsio/status/1520356430800617472

Expensify got 4 million request _per second_ out of a custom SQLite-based setup (on a huge machine, but still): https://blog.expensify.com/2018/01/08/scaling-sqlite-to-4m-q...

I would call those serious production apps.

I specifically cited concurrent writes. SQLite is fantastic for read-only data. Those websites are using SQLite for what it's designed for. But, if they were trying to use SQLite to do even dozens of writes per second, it would crash and burn.

Well, that claim is fairly easy to test. So I did: https://gist.github.com/markuswustenberg/0030360d7d4679f2656...

Nothing fancy, just a small table that holds some blog posts with an ID, a title, some content, and a creation timestamp.

I ran the benchmark with and without WAL enabled on my Macbook Air (2020, M1) with some SSD drive inside. Results:

  $ make benchmark 
  go test -bench=.
  goos: darwin
  goarch: arm64
  pkg: sqlite
  BenchmarkWriteBlogPost/write_blog_post_without_WAL-8              6441     191735 ns/op
  BenchmarkWriteBlogPost/write_blog_post_with_WAL-8               102559      11205 ns/op
  PASS
  ok   sqlite 3.725s

That's around 89k writes per second in parallel on all available cores with WAL enabled. I know this is a trivial setup, but adjust to your liking. You'll find that SQLite probably doesn't crash and burn with dozens of writes.

This is single-threaded, which a production app would not be. That's when things get scary with SQLite.

Yes, I was incorrect when I said SQLite couldn't handle lots of writes quickly; what I should have said is that it can't handle lots of writes from multiple threads quickly.

I'm not understanding this. Wouldn't the lower bound on performance be the single thread? Any worse performance and you queue the write requests to match the single thread performance.

No, because of SQLite's locking mechanism. It uses filesytem-level constructs to support concurrency, and this gets dicey, quickly.

As my test above shows, use a mutex for writes in Go directly and whatever locking performance problems SQLite exhibits disappear. I suspect because of the polling used in SQLite as described elsewhere on this page.

Okay, fair enough. I thought my previous test was parallel as well, but I was mistaken.

Here's one for 64 parallel writes: https://gist.github.com/markuswustenberg/f35ab7e191137dca5f7...

  $ make benchmark
  go test -bench=.
  goos: darwin
  goarch: arm64
  pkg: sqlite
  BenchmarkWriteBlogPost/write_blog_post_without_WAL-8               100   16317415 ns/op
  BenchmarkWriteBlogPost/write_blog_post_with_WAL-8                 1196    1198679 ns/op
  BenchmarkWriteBlogPost/write_blog_post_with_WAL_and_Go_mutex-8             58455      17461 ns/op
  PASS
  ok   sqlite 4.557s

1e9/1198679 = 834 writes per second is still far from crash and burn territory when using just WAL mode.

Of course, it gets more interesting when there are also concurrent readers, as people are trying out elsewhere in the discussion. But the point still stands: it can handle way, waaaay better than "fives of authors" in a not-at-all scary way.

You should duplicate this test against postgres inserts. I did the same thing for sqlite; in go using a mutex on inserts, and sqlite still out performs postgres on inserts (on the same machine).

Reply on news.ycombinator.com