Sunbelt Computer Software

PL/B Language Development and Support

Home - SIEVE: an Efficient Turn-Key Eviction Algorithm for Web Caches

SIEVE

How does SIEVE work?

Key Features

Simplicity

SIEVE is easy to implement and can be easily integrated into existing systems.

Cache library	Language	Lines	Hour of work
groupcache	Golang	21	<1
mnemonist	Javascript	12	1
lru-rs	Rust	16	1
lru-dict	Python + C	21	<1

Efficiency

SIEVE achieves state-of-the-art efficiency on skewed workloads.

*Miss ratio on a Zipfian dataset (α=1).

Cache Primitive

SIEVE can facilitate the design of advanced eviction algorithms.

Adoption

Large systems:

Immudb

TiDB

Pelikan

Nyrkiö

SkiftOS

DragonFly

DNSCrypt-proxy

encrypted-dns-resolver

PostgREST

Cache libraries:

SieveCache

golang-fifo

js-sieve

rust-sieve-cache

go-sieve

sieve_cache (Ruby)

zig-sieve (Zig)

sieve (Swift)

sieve-cache-d (D)

sieve-cache-cpp (C++)

sieve (JavaScript)

sieve (Elixir)

sieve (Nim)

SIEVE (C#)

sieve-cache (Java)

sieve (Python)

sieve-cache-in-rust

sieve-cache (JavaScript)

sieve (typescrpt) samber/hot (Go)

Let us know if you want to be featured here!

Media coverage:

TL;DR

Talk at IEPG - Internet Engineering and Planning Group

We would love to hear your experience!

Acknowledgement

Sponsors:
Carnegie Mellon University Parallel Data Laboratory Emory University SimBioSys Lab
Meta, Google, AWS, Microsoft, CloudLab, NSF, Sloan Foundation and VMware

Dataset source:
Meta, Tencent, Wikimedia, Twitter