Skip to content

Hit ratio

Traditional traces

Zipf

zipf

S3

This trace is described as "disk read accesses initiated by a large commercial search engine in response to various web search requests.".

s3

DS1

This trace is described as "a database server running at a commercial site running an ERP application on top of a commercial database.".

ds1

P3

The trace P3 was collected from workstations running Windows NT by using Vtrace which captures disk operations through the use of device filters.

p3

P8

The trace P8 was collected from workstations running Windows NT by using Vtrace which captures disk operations through the use of device filters.

p8

LOOP

This trace demonstrates a looping access pattern.

loop

OLTP

This trace is described as "references to a CODASYL database for a one-hour period.".

oltp

Modern traces

WikiCDN

This trace was collected from Wikimedia's CDN system in 2019.

wikicdn

AlibabaBlock

This trace was collected from a cluster in production of the elastic block service of Alibaba Cloud (i.e. storage for virtual disks) in 2020.

alibabablock

Twitter

This trace was collected from Twitter's in-memory caching (Twemcache/Pelikan) clusters in 2020.

twitter

Conclusion

S3-FIFO (otter) is inferior to W-TinyLFU (theine) on lfu friendly traces (databases, search, analytics) and has a greater or equal hit ratio on web traces. But it is still worth recognizing that W-TinyLFU is superior to S3-FIFO as a general-purpose eviction policy.

In summary, we have that S3-FIFO is competitive with W-TinyLFU and ARC. Also, it provides a substantial improvement to LRU across a variety of traces.