Investigation: Why is SQS so slow?

[luu]

https://blog.afoolishmanifesto.com/posts/investigation-why-sqs-slow/

[Lazare]

A better title might be: "The Perl library Furl is horrible; use something vaguely functional instead if you need to make HTTP/S requests using Perl."

Good tip, if I was using Furl (or even Perl), but utterly unrelated to SQS, and not obviously generalisable to other libraries or languages.

[tyingq → Lazare]

It doesn't even appear to be the obvious library to use to access SQS from Perl.

Odd article.

[vacri → Lazare]

It's not the best title, but it fits. The initial symptom that kicked off the investigation was a slowdown talking to SQS.

[eknkc]

Up next: Why is desktop linux so unresponsive? A deep dive into my broken keyboard.

[akerl_]

This feels more like I'm reading the intro to a hypothetical blog post. What was the thing that Furl did that was bad? How does Curl do it better? Is there anything SQS-specific about the issue, or was the initial setup about queuing systems just extraneous info?

[czardoz]

tl;dr: The author was seeing some network issues. Switching the HTTP client library fixed them.

[yeukhon]

Good thing I read the comment before I read the article, because I couldn't really follow the post... I am not alone. Anyway, I am glad author figured it out!

So I was looking up "furl" [1] and the first thing returned was a Python library called "furl". But the actual "Furl" in the blog post is a Perl library [2]. Open source project owners, please hear me out... can we please avoid name collision?

[1]: https://github.com/gruns/furl [2]: https://github.com/tokuhirom/Furl

[andrewstuart]

I didn't carefully read every word but a skim didn't seem to answer the question in the headline.

And I'm not interested enough from what I saw to read every word. Something about curl?

[magoon]

Extremely interested; didn’t finish once in the author got into the weeds with sysctl and furl.