---
title: "Comment by Vũ Nguyễn on \"Scalable Bloom Filter \" | Hashnode"
description: "Cho em hỏi tại sao không dùng BloomFilter.approximateElementCount mà lại dùng HyperLogLog vậy ạ. https://guava.dev/releases/33.4.8-android/api/docs/com/google/c"
url: "https://hashnode.com/posts/scalable-bloom-filter/6a81d052b0afa681ea4aa0b6/comment/6a8308181a7a118663ff18a3"
---

[Discussion](https://hashnode.com/posts/scalable-bloom-filter/6a81d052b0afa681ea4aa0b6)

Comment [Scalable Bloom Filter](https://hashnode.com/posts/scalable-bloom-filter/6a81d052b0afa681ea4aa0b6)

[Vũ Nguyễn](https://hashnode.com/@eaglevee)

[Aug 17](https://hashnode.com/posts/scalable-bloom-filter/6a81d052b0afa681ea4aa0b6/comment/6a8308181a7a118663ff18a3)

Cho em hỏi tại sao không dùng BloomFilter.approximateElementCount mà lại dùng HyperLogLog vậy ạ.

[guava.dev/releases/33.4.8-android/api/docs/com/go…](https://guava.dev/releases/33.4.8-android/api/docs/com/google/common/hash/BloomFilter.html)

[nguyen nam](https://hashnode.com/@namnt1410)

[Aug 18](https://hashnode.com/posts/scalable-bloom-filter/6a81d052b0afa681ea4aa0b6/comment/6a8308181a7a118663ff18a3#reply-6a8431234f8c8093b645c4af)

[Vũ Nguyễn](https://hashnode.com/@eaglevee "Vũ Nguyễn") do guava lưu mem không lưu vào disk, pod off là mất

[Vũ Nguyễn](https://hashnode.com/@eaglevee)

[Aug 20](https://hashnode.com/posts/scalable-bloom-filter/6a81d052b0afa681ea4aa0b6/comment/6a8308181a7a118663ff18a3#reply-6a873b518f763e38fd1f1c96)

Vậy nếu write stream guava vào disk/redis và rebuild lúc restart thì có tốt hơn dùng HLL không anh, đỡ phải dùng nhiều tầng

[nguyen nam](https://hashnode.com/@namnt1410)

[Aug 30](https://hashnode.com/posts/scalable-bloom-filter/6a81d052b0afa681ea4aa0b6/comment/6a8308181a7a118663ff18a3#reply-6a93746b526c4de3440ca876)

Mỗi lần start 1 pod lại rebuild thì nặng nếu bloom filter lớn, client phải chờ pod rebuild xong mới query được