Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Speaker Deck
PRO
Sign in
Sign up for free
Distributed Elixir
Maciej Kaszubowski
July 07, 2018
Programming
0
55
Distributed Elixir
Presentation about some of the tools for distributed programming in Elixir
Maciej Kaszubowski
July 07, 2018
Tweet
Share
More Decks by Maciej Kaszubowski
See All by Maciej Kaszubowski
Error-free Elixir
mkaszubowski
0
110
Modular Design in Elixir (ElixirConf EU 2019)
mkaszubowski
2
430
The Big Ball of Nouns
mkaszubowski
0
78
Modular Design in Elixir
mkaszubowski
1
210
Our three years with Elixir
mkaszubowski
0
95
Concurrency Basics for Elixir
mkaszubowski
0
56
Software Architecture
mkaszubowski
0
53
Let it crash - fault tolerance in Elixir/OTP
mkaszubowski
0
200
CRDTs - The science behind Phoenix Presence
mkaszubowski
2
160
Other Decks in Programming
See All in Programming
GraphQL+KMM開発でわかったこと / What we learned from GraphQL+KMM development
kubode
0
130
バンドル最適化マニアクス at tfconf
mizchi
3
2.2k
New Relicを使った Observabilityの実現方法と活用例 / gocon 2022 spring after talk
budougumi0617
0
1.1k
byte列のbit表現を得るencodingライブラリ作った
convto
1
130
tfcon2022_Web3Dひとめぐり.pdf
emadurandal
0
960
dbtとBigQueryで始めるData Vault入門
kazk1018
0
180
Kueue入門/Kueue Introduction
bells17
0
510
Micro Frontends with Module Federation: Beyond the Basics @codecrafts2022
manfredsteyer
PRO
0
100
Jakarta EE 10 is Coming Your Way
ivargrimstad
0
2.6k
Swift Concurrencyによる安全で快適な非同期処理
tattn
2
320
LOWYAの信頼性向上とNew Relic
kazumax55
4
350
Loom is Blooming
josepaumard
3
540
Featured
See All Featured
VelocityConf: Rendering Performance Case Studies
addyosmani
316
22k
What's new in Ruby 2.0
geeforr
336
30k
Designing on Purpose - Digital PM Summit 2013
jponch
106
5.6k
How STYLIGHT went responsive
nonsquared
85
3.9k
Code Reviewing Like a Champion
maltzj
506
37k
From Idea to $5000 a Month in 5 Months
shpigford
372
44k
Happy Clients
brianwarren
89
5.5k
GitHub's CSS Performance
jonrohan
1020
410k
Making the Leap to Tech Lead
cromwellryan
113
6.9k
JavaScript: Past, Present, and Future - NDC Porto 2020
reverentgeek
37
3.2k
Build your cross-platform service in a week with App Engine
jlugia
219
17k
The Language of Interfaces
destraynor
148
20k
Transcript
It’s scary out there
Organisational Matters
None
We’re 1 year old!
Summer break (probably)
We’re looking for speakers!
It’s scary out there Distributed Systems in Elixir Poznań Elixir
Meetup #8
None
Pid 1 Pid 2
Pid 1 Pid 2 Node A Node B
The basics
iex --name n1@127.0.0.1 --cookie cookie -S mix
Node.connect(:’n1@127.0.0.1')
(DEMO)
#PID<0.94.0>
#PID<0.94.0> node identifier (relative to current node)
#PID<0.94.0> node identifier (relative to current node) 0 =a local
process
#PID<0.94.0> Process id node identifier (relative to current node)
How does it work?
Pid 1 Node A Pid 2 Node B
Pid 1 Node A Pid 2 Node B TCP Connection
send(pid2, msg) Pid 1 Node A Pid 2 Node B
TCP Connection
send(pid2, msg) Pid 1 Node A Pid 2 Node B
destination_node = node(pid) TCP Connection
send(pid2, msg) Pid 1 Node A Pid 2 Node B
destination_node = node(pid) :erlang.term_to_binary(msg) TCP Connection
send(pid2, msg) Pid 1 Node A Pid 2 Node B
destination_node = node(pid) :erlang.term_to_binary(msg) TCP Connection
send(pid2, msg) Pid 1 Node A Pid 2 Node B
destination_node = node(pid) :erlang.term_to_binary(msg) TCP Connection :erlang.binary_to_term(encode)
send(pid2, msg) Pid 1 Node A receive msg Pid 2
Node B destination_node = node(pid) :erlang.term_to_binary(msg) TCP Connection :erlang.binary_to_term(encode)
Distributed Systems?
Distributed Systems? Solved!
Well, not exactly…
Difficulties
Node A Node B
Node A Node B Node C
Node A Node B Node C Node D
None
A lot of messages
us-east-1 us-west-2
8 fallacies of distributed computing
fallacies of distributed computing 1. The network is reliable 2.
Latency is zero 3. Bandwidth is infinite 4. The network is secure 5. Topology doesn’t change 6. The is one administrator 7. Transport cost is zero 8. The network is homogenous
CAP THEOREM
CAP THEOREM us-west-2 us-east-1
CAP THEOREM us-west-2 us-east-1 Set X=5
CAP THEOREM us-west-2 us-east-1 Set X=5 Read X
CAP THEOREM us-west-2 us-east-1 Set X=5 Set X = 7
Consistency or Availability (under network partition)
Consistency or Speed In practice
Guarantees
Pid 1 Pid 2 Pid3 Guarantees m1, m2, m3 m4,
m5, m6 send(pid2, m1) send(pid2, m2) send(pid2, m3) send(pid2, m4) send(pid2, m5) send(pid2, m6)
Pid 1 Pid 2 Pid3 Guarantees m1, m2, m3 m4,
m5, m6 send(pid2, m1) send(pid2, m2) send(pid2, m3) send(pid2, m4) send(pid2, m5) send(pid2, m6) Ordering between two processes is preserved
Pid 1 Pid 2 Pid3 Guarantees m4, m5, m6 send(pid2,
m1) send(pid2, m2) send(pid2, m3) send(pid2, m4) send(pid2, m5) send(pid2, m6) m1, m2, m3 Delivery is not guaranteed
Pid 1 Pid 2 Pid3 Guarantees m1, m2, m3 m4,
m5, m6 send(pid2, m1) send(pid2, m2) send(pid2, m3) send(pid2, m4) send(pid2, m5) send(pid2, m6) Ordering between different processes is not guaranteed
[m1, m2, m3, m4, m5, m6]
[m1, m2, m3, m4, m5, m6] [m4, m5, m6, m1,
m2, m3]
[m1, m2, m3, m4, m5, m6] [m4, m5, m6, m1,
m2, m3] [m1, m4, m2, m5, m3, m6]
[m1, m2, m3, m4, m5, m6] [m4, m5, m6, m1,
m2, m3] [m1, m4, m2, m5, m3, m6] [m1, m2, m3]
[m1, m2, m3, m4, m5, m6] [m4, m5, m6, m1,
m2, m3] [m1, m4, m2, m5, m3, m6] [m1, m2, m3] [m1, m3, m5, m6]
[m1, m2, m3, m4, m5, m6] [m4, m5, m6, m1,
m2, m3] [m1, m4, m2, m5, m3, m6] [m1, m2, m3] [m1, m3, m5, m6] []
[m1, m2, m3, m4, m5, m6] [m4, m5, m6, m1,
m2, m3] [m1, m4, m2, m5, m3, m6] [m1, m2, m3] [m1, m3, m5, m6] [] [m1, m3, m2, m4, m5, m6]
[m1, m2, m3, m4, m5, m6] [m4, m5, m6, m1,
m2, m3] [m1, m4, m2, m5, m3, m6] [m1, m2, m3] [m1, m3, m5, m6] [] [m1, m3, m2, m4, m5, m6] [M3, M3]
Phoenix Request A User Logged In
Phoenix Request A Phoenix Request B User Logged In User
Logged OUT
Phoenix Request A Phoenix Request B User Logged In User
Logged OUT This Can arrive first
Unfortunately, things tend to work fine locally
The Tools
:global
Pid 1 Node A Node B Pid 2
Pid 1 Node A Node B Pid 2 :global.register_name(“global”, self())
Pid 1 Node A Node B Pid 2 :global.register_name(“global”, self())
Register PId1 as “global”
Pid 1 Node A Node B Pid 2 :global.register_name(“global”, self())
Register PId1 as “global” Sure
Pid 1 Node A Node B Pid 2 :global.register_name(“global”, self())
Register PId1 as “global” Sure :global.whereis_name(“global”) = pid1
Pid 1 Node A Node B Pid 2 :global.register_name(“global”, self())
:global.register_name(“global”, self()) ?
(DEMO)
:global • single process registration (if everything works OK) •
Favours availability over consistency • Information stored locally (reading is fast) • Registration is blocking (may be slow)
:PG2
Pid1 Pid3 Pid2 [] [] []
Pid1 Pid3 Pid2 :pg2.create(“my_group”) [] [] []
Pid1 Pid3 Pid2 [] [] [] join join :pg2.join(“my_group”, self()
Pid1 Pid3 Pid2 [] [pid1] [] Monitor Monitor :pg2.join(“my_group”, self()
Pid1 Pid3 Pid2 [pid1] [pid1] [pid1] Monitor Monitor :pg2.join(“my_group”, self()
Pid1 Pid3 Pid2 [pid1] [pid1] [pid1]
Pid1 Pid3 Pid2 :pg2.join(“my_group”, self() [pid1] [pid1, pid2] [pid1]
Pid1 Pid3 Pid2 join :pg2.join(“my_group”, self() join [pid1, pid2] [pid1,
pid2] [pid1, pid2]
Pid1 Pid3 Pid2 [pid1] [pid2] [pid1]
Pid1 Pid3 Pid2 [pid1] [pid2] [pid1]
Pid1 Pid3 Pid2 [pid1] [pid2] [pid1]
Pid1 Pid3 Pid2 [pid1, pid2] [pid1, pid2] [pid1, pid2]
It will heal, but the state in inconsistent for some
time
What does it matter?
Node A Pg2 Pg2 Pg2 Node B Node C
Node A Pg2 Pg2 Pg2 Node B Node C Phoenix
Channels
Node A Pg2 Pg2 Pg2 Node B Node C Phoenix
Presence
Node A Pg2 Pg2 Pg2 Node B Node C Phoenix
Channels
:pg2 • Process groups • Favours availability over consistency •
Information stored locally (reading is fast) • Registration is blocking (may be slow)
Strongly consistent Solutions
Strongly consistent Solutions • Consensus - Raft, Paxos, ZAB •
Two-phase commit/THree-phase commit (2PC/3PC) • Read/Write quorums • Single database as a source of truth
Summary
Distributed Systems
Well, not exactly…
Asynchronous messages Distributed systems are all about
Really, there’s no magic
Just asynchronous messages between nodes
Just asynchronous messages between nodes & node failures
Just asynchronous messages between nodes & node failures & Communication
failures
Just asynchronous messages between nodes & node failures & Communication
failures & Network partitions
Tradeoffs Distributed systems are all about
Where to go next
Worth looking at • Riak_core • RAFT • Two-Phase Commit
(2PC) / Three-Phase Commit (3PC) • CRDTs • LASP and Partisan
Free online (click!) Elixir / Erlang
Free PDF (Click!) Distributed Systems
Theory (The hard stuff)
• https://raft.github.io/ (Raft Consensus) • http://learnyousomeerlang.com/distribunomicon • https://www.rgoarchitects.com/Files/fallacies.pdf (Fallacies of
distributed computing) • https://dzone.com/articles/better-explaining-cap-theorem (CAP Theorem) • https://medium.com/learn-elixir/message-order-and-delivery-guarantees-in-elixir- erlang-9350a3ea7541 (Elixir message delivery guarantees) • https://lasp-lang.readme.io/ (LASP) • https://arxiv.org/pdf/1802.02652.pdf (Partisan Paper) • https://bravenewgeek.com/tag/three-phase-commit/ (3PC)
We’re looking for speakers!
Thank You! Poznań Elixir Meetup #8