Patch series "Multi-Gen LRU Framework", v14.
What's new
==========
1. OpenWrt, in addition to Android, Arch Linux Zen, Armbian, ChromeOS,
Liquorix, post-factum and XanMod, is now shipping MGLRU on 5.15.
2. Fixed long-tailed direct reclaim latency seen on high-memory (TBs)
machines. The old direct reclaim backoff, which tries to enforce a
minimum fairness among all eligible memcgs, over-swapped by about
(total_mem>>DEF_PRIORITY)-nr_to_reclaim. The new backoff, which
pulls the plug on swapping once the target is met, trades some
fairness for curtailed latency:
https://lore.kernel.org/r/20220918080010.2920238-10-yuzhao@google.com/
3. Fixed minior build warnings and conflicts. More comments and nits.
TLDR
====
The current page reclaim is too expensive in terms of CPU usage and it
often makes poor choices about what to evict. This patchset offers an
alternative solution that is performant, versatile and
straightforward.
Patchset overview
=================
The design and implementation overview is in patch 14:
https://lore.kernel.org/r/20220918080010.2920238-15-yuzhao@google.com/
01. mm: x86, arm64: add arch_has_hw_pte_young()
02. mm: x86: add CONFIG_ARCH_HAS_NONLEAF_PMD_YOUNG
Take advantage of hardware features when trying to clear the accessed
bit in many PTEs.
03. mm/vmscan.c: refactor shrink_node()
04. Revert "include/linux/mm_inline.h: fold __update_lru_size() into
its sole caller"
Minor refactors to improve readability for the following patches.
05. mm: multi-gen LRU: groundwork
Adds the basic data structure and the functions that insert pages to
and remove pages from the multi-gen LRU (MGLRU) lists.
06. mm: multi-gen LRU: minimal implementation
A minimal implementation without optimizations.
07. mm: multi-gen LRU: exploit locality in rmap
Exploits spatial locality to improve efficiency when using the rmap.
08. mm: multi-gen LRU: support page table walks
Further exploits spatial locality by optionally scanning page tables.
09. mm: multi-gen LRU: optimize multiple memcgs
Optimizes the overall performance for multiple memcgs running mixed
types of workloads.
10. mm: multi-gen LRU: kill switch
Adds a kill switch to enable or disable MGLRU at runtime.
11. mm: multi-gen LRU: thrashing prevention
12. mm: multi-gen LRU: debugfs interface
Provide userspace with features like thrashing prevention, working set
estimation and proactive reclaim.
13. mm: multi-gen LRU: admin guide
14. mm: multi-gen LRU: design doc
Add an admin guide and a design doc.
Benchmark results
=================
Independent lab results
-----------------------
Based on the popularity of searches [01] and the memory usage in
Google's public cloud, the most popular open-source memory-hungry
applications, in alphabetical order, are:
Apache Cassandra Memcached
Apache Hadoop MongoDB
Apache Spark PostgreSQL
MariaDB (MySQL) Redis
An independent lab evaluated MGLRU with the most widely used benchmark
suites for the above applications. They posted 960 data points along
with kernel metrics and perf profiles collected over more than 500
hours of total benchmark time. Their final reports show that, with 95%
confidence intervals (CIs), the above applications all performed
significantly better for at least part of their benchmark matrices.
On 5.14:
1. Apache Spark [02] took 95% CIs [9.28, 11.19]% and [12.20, 14.93]%
less wall time to sort three billion random integers, respectively,
under the medium- and the high-concurrency conditions, when
overcommitting memory. There were no statistically significant
changes in wall time for the rest of the benchmark matrix.
2. MariaDB [03] achieved 95% CIs [5.24, 10.71]% and [20.22, 25.97]%
more transactions per minute (TPM), respectively, under the medium-
and the high-concurrency conditions, when overcommitting memory.
There were no statistically significant changes in TPM for the rest
of the benchmark matrix.
3. Memcached [04] achieved 95% CIs [23.54, 32.25]%, [20.76, 41.61]%
and [21.59, 30.02]% more operations per second (OPS), respectively,
for sequential access, random access and Gaussian (distribution)
access, when THP=always; 95% CIs [13.85, 15.97]% and
[23.94, 29.92]% more OPS, respectively, for random access and
Gaussian access, when THP=never. There were no statistically
significant changes in OPS for the rest of the benchmark matrix.
4. MongoDB [05] achieved 95% CIs [2.23, 3.44]%, [6.97, 9.73]% and
[2.16, 3.55]% more operations per second (OPS), respectively, for
exponential (distribution) access, random access and Zipfian
(distribution) access, when underutilizing memory; 95% CIs
[8.83, 10.03]%, [21.12, 23.14]% and [5.53, 6.46]% more OPS,
respectively, for exponential access, random access and Zipfian
access, when overcommitting memory.
On 5.15:
5. Apache Cassandra [06] achieved 95% CIs [1.06, 4.10]%, [1.94, 5.43]%
and [4.11, 7.50]% more operations per second (OPS), respectively,
for exponential (distribution) access, random access and Zipfian
(distribution) access, when swap was off; 95% CIs [0.50, 2.60]%,
[6.51, 8.77]% and [3.29, 6.75]% more OPS, respectively, for
exponential access, random access and Zipfian access, when swap was
on.
6. Apache Hadoop [07] took 95% CIs [5.31, 9.69]% and [2.02, 7.86]%
less average wall time to finish twelve parallel TeraSort jobs,
respectively, under the medium- and the high-concurrency
conditions, when swap was on. There were no statistically
significant changes in average wall time for the rest of the
benchmark matrix.
7. PostgreSQL [08] achieved 95% CI [1.75, 6.42]% more transactions per
minute (TPM) under the high-concurrency condition, when swap was
off; 95% CIs [12.82, 18.69]% and [22.70, 46.86]% more TPM,
respectively, under the medium- and the high-concurrency
conditions, when swap was on. There were no statistically
significant changes in TPM for the rest of the benchmark matrix.
8. Redis [09] achieved 95% CIs [0.58, 5.94]%, [6.55, 14.58]% and
[11.47, 19.36]% more total operations per second (OPS),
respectively, for sequential access, random access and Gaussian
(distribution) access, when THP=always; 95% CIs [1.27, 3.54]%,
[10.11, 14.81]% and [8.75, 13.64]% more total OPS, respectively,
for sequential access, random access and Gaussian access, when
THP=never.
Our lab results
---------------
To supplement the above results, we ran the following benchmark suites
on 5.16-rc7 and found no regressions [10].
fs_fio_bench_hdd_mq pft
fs_lmbench pgsql-hammerdb
fs_parallelio redis
fs_postmark stream
hackbench sysbenchthread
kernbench tpcc_spark
memcached unixbench
multichase vm-scalability
mutilate will-it-scale
nginx
[01] https://trends.google.com
[02] https://lore.kernel.org/r/20211102002002.92051-1-bot@edi.works/
[03] https://lore.kernel.org/r/20211009054315.47073-1-bot@edi.works/
[04] https://lore.kernel.org/r/20211021194103.65648-1-bot@edi.works/
[05] https://lore.kernel.org/r/20211109021346.50266-1-bot@edi.works/
[06] https://lore.kernel.org/r/20211202062806.80365-1-bot@edi.works/
[07] https://lore.kernel.org/r/20211209072416.33606-1-bot@edi.works/
[08] https://lore.kernel.org/r/20211218071041.24077-1-bot@edi.works/
[09] https://lore.kernel.org/r/20211122053248.57311-1-bot@edi.works/
[10] https://lore.kernel.org/r/20220104202247.2903702-1-yuzhao@google.com/
Read-world applications
=======================
Third-party testimonials
------------------------
Konstantin reported [11]:
I have Archlinux with 8G RAM + zswap + swap. While developing, I
have lots of apps opened such as multiple LSP-servers for different
langs, chats, two browsers, etc... Usually, my system gets quickly
to a point of SWAP-storms, where I have to kill LSP-servers,
restart browsers to free memory, etc, otherwise the system lags
heavily and is barely usable.
1.5 day ago I migrated from 5.11.15 kernel to 5.12 + the LRU
patchset, and I started up by opening lots of apps to create memory
pressure, and worked for a day like this. Till now I had not a
single SWAP-storm, and mind you I got 3.4G in SWAP. I was never
getting to the point of 3G in SWAP before without a single
SWAP-storm.
Vaibhav from IBM reported [12]:
In a synthetic MongoDB Benchmark, seeing an average of ~19%
throughput improvement on POWER10(Radix MMU + 64K Page Size) with
MGLRU patches on top of 5.16 kernel for MongoDB + YCSB across
three different request distributions, namely, Exponential, Uniform
and Zipfan.
Shuang from U of Rochester reported [13]:
With the MGLRU, fio achieved 95% CIs [38.95, 40.26]%, [4.12, 6.64]%
and [9.26, 10.36]% higher throughput, respectively, for random
access, Zipfian (distribution) access and Gaussian (distribution)
access, when the average number of jobs per CPU is 1; 95% CIs
[42.32, 49.15]%, [9.44, 9.89]% and [20.99, 22.86]% higher
throughput, respectively, for random access, Zipfian access and
Gaussian access, when the average number of jobs per CPU is 2.
Daniel from Michigan Tech reported [14]:
With Memcached allocating ~100GB of byte-addressable Optante,
performance improvement in terms of throughput (measured as queries
per second) was about 10% for a series of workloads.
Large-scale deployments
-----------------------
We've rolled out MGLRU to tens of millions of ChromeOS users and
about a million Android users. Google's fleetwide profiling [15] shows
an overall 40% decrease in kswapd CPU usage, in addition to
improvements in other UX metrics, e.g., an 85% decrease in the number
of low-memory kills at the 75th percentile and an 18% decrease in
app launch time at the 50th percentile.
The downstream kernels that have been using MGLRU include:
1. Android [16]
2. Arch Linux Zen [17]
3. Armbian [18]
4. ChromeOS [19]
5. Liquorix [20]
6. OpenWrt [21]
7. post-factum [22]
8. XanMod [23]
[11] https://lore.kernel.org/r/140226722f2032c86301fbd326d91baefe3d7d23.camel@yandex.ru/
[12] https://lore.kernel.org/r/87czj3mux0.fsf@vajain21.in.ibm.com/
[13] https://lore.kernel.org/r/20220105024423.26409-1-szhai2@cs.rochester.edu/
[14] https://lore.kernel.org/r/CA+4-3vksGvKd18FgRinxhqHetBS1hQekJE2gwco8Ja-bJWKtFw@mail.gmail.com/
[15] https://dl.acm.org/doi/10.1145/2749469.2750392
[16] https://android.com
[17] https://archlinux.org
[18] https://armbian.com
[19] https://chromium.org
[20] https://liquorix.net
[21] https://openwrt.org
[22] https://codeberg.org/pf-kernel
[23] https://xanmod.org
Summary
=======
The facts are:
1. The independent lab results and the real-world applications
indicate substantial improvements; there are no known regressions.
2. Thrashing prevention, working set estimation and proactive reclaim
work out of the box; there are no equivalent solutions.
3. There is a lot of new code; no smaller changes have been
demonstrated similar effects.
Our options, accordingly, are:
1. Given the amount of evidence, the reported improvements will likely
materialize for a wide range of workloads.
2. Gauging the interest from the past discussions, the new features
will likely be put to use for both personal computers and data
centers.
3. Based on Google's track record, the new code will likely be well
maintained in the long term. It'd be more difficult if not
impossible to achieve similar effects with other approaches.
This patch (of 14):
Some architectures automatically set the accessed bit in PTEs, e.g., x86
and arm64 v8.2. On architectures that do not have this capability,
clearing the accessed bit in a PTE usually triggers a page fault following
the TLB miss of this PTE (to emulate the accessed bit).
Being aware of this capability can help make better decisions, e.g.,
whether to spread the work out over a period of time to reduce bursty page
faults when trying to clear the accessed bit in many PTEs.
Note that theoretically this capability can be unreliable, e.g.,
hotplugged CPUs might be different from builtin ones. Therefore it should
not be used in architecture-independent code that involves correctness,
e.g., to determine whether TLB flushes are required (in combination with
the accessed bit).
Link: https://lkml.kernel.org/r/20220918080010.2920238-1-yuzhao@google.com
Link: https://lkml.kernel.org/r/20220918080010.2920238-2-yuzhao@google.com
Signed-off-by: Yu Zhao <yuzhao@google.com>
Reviewed-by: Barry Song <baohua@kernel.org>
Acked-by: Brian Geffon <bgeffon@google.com>
Acked-by: Jan Alexander Steffens (heftig) <heftig@archlinux.org>
Acked-by: Oleksandr Natalenko <oleksandr@natalenko.name>
Acked-by: Steven Barrett <steven@liquorix.net>
Acked-by: Suleiman Souhlal <suleiman@google.com>
Acked-by: Will Deacon <will@kernel.org>
Tested-by: Daniel Byrne <djbyrne@mtu.edu>
Tested-by: Donald Carr <d@chaos-reins.com>
Tested-by: Holger Hoffstätte <holger@applied-asynchrony.com>
Tested-by: Konstantin Kharlamov <Hi-Angel@yandex.ru>
Tested-by: Shuang Zhai <szhai2@cs.rochester.edu>
Tested-by: Sofia Trinh <sofia.trinh@edi.works>
Tested-by: Vaibhav Jain <vaibhav@linux.ibm.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Hillf Danton <hdanton@sina.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: linux-arm-kernel@lists.infradead.org
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Michael Larabel <Michael@MichaelLarabel.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Qi Zheng <zhengqi.arch@bytedance.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>