4 ms·
I also thought cache poisoning means putting a bad response into a cache... I agree. The cached responses mentioned in this article are perfectly valid; they'
by cperciva 11y ago
I also thought cache poisoning means putting a bad response into a cache...
I agree. The cached responses mentioned in this article are perfectly valid; they're simply not very useful.
Rather than cache poisoning, I'd call this cache constipation.
- jacquesm 11y agoI can't imagine PHK making such a basic mistake, who wrote the post?
- cperciva 11y agoThe HTML header says <meta name="author" content="Dridi Boukelmoune"> who according to crunchbase is a "Technical Presales Engineer" at Varnish.
- dridi 11y agoSorry for not misunderstanding. I'm not saying that PHK made a mistake, he picked the simple solution (which is very often the case). But LRU being too simple, you need to do more to fight again cache pollution. Segmenting your storage is the solution I suggest in the blog post.
- dridi 11y agoLet's call that cache pollution or flooding attacks. In the end, I consider it poisonous because I don't want to evict relevant content for junk (especially from the long tail). I do mention that the RFC says that 404 responses can be cached, and that according to my own cache policy, I disagree. It depends on your goals.