Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
Reliability of Distributed Systems
Search
Piyush Verma
June 22, 2019
Technology
0
170
Reliability of Distributed Systems
Piyush Verma
June 22, 2019
Tweet
Share
More Decks by Piyush Verma
See All by Piyush Verma
SLOs that Lie
meson10
0
84
Doing SRE the right way - 2
meson10
0
150
Doing SRE the right way
meson10
0
930
Observability and Control Theory
meson10
1
920
Reliability
meson10
0
91
My TLS was broken
meson10
0
74
Technology that builds Organizations
meson10
0
71
Namespace.go
meson10
0
81
Cgroups and Namespaces in Linux
meson10
0
280
Other Decks in Technology
See All in Technology
Application Signalsで始めるSLO ユーザー満足度を数値化する第一歩
niftycorp
PRO
2
160
Vue.js、Nuxtの機能を使い、 大量のコピペコードをリファクタリングする
igayamaguchi
3
1.4k
本番のトラフィック量でHudiを検証して見えてきた課題
joker1007
2
270
Oracle Base Database Service 技術詳細
oracle4engineer
PRO
5
48k
Deep dive into Nuxt Server Components
wattanx
1
1.3k
AWS Lambda と Amazon SQS で「わかった気になれる」FreeRTOS 入門
soracom
PRO
2
150
RSGT Walk Through
kawaguti
PRO
2
400
テクニカルライターのチームで「目標」をどう決めたか / MVV for a Team of Technical Writers
lycorptech_jp
PRO
3
160
Amazon Managed Grafana で AWS IoT TwinMaker によるデジタルツインアプリケーションを動かしてみた
wakatsuki
0
120
人工衛星開発のための C2A フレームワークとその開発体験
sksat
0
110
Amazon ECS & AWS Fargate 今昔物語 / past and present stories of Amazon ECS and AWS Fargate
iselegant
19
4.1k
v-modelの歩みを振り返る
bengo4com
5
2.4k
Featured
See All Featured
Building Adaptive Systems
keathley
38
2.2k
Embracing the Ebb and Flow
colly
84
4.4k
Why You Should Never Use an ORM
jnunemaker
PRO
53
9k
Put a Button on it: Removing Barriers to Going Fast.
kastner
58
3.5k
The Success of Rails: Ensuring Growth for the Next 100 Years
eileencodes
43
6.6k
How STYLIGHT went responsive
nonsquared
95
5.1k
Imperfection Machines: The Place of Print at Facebook
scottboms
264
13k
Producing Creativity
orderedlist
PRO
341
39k
Typedesign – Prime Four
hannesfritz
39
2.4k
Exploring the Power of Turbo Streams & Action Cable | RailsConf2023
kevinliebholz
26
4.1k
Refactoring Trust on Your Teams (GOTO; Chicago 2020)
rmw
31
2.6k
jQuery: Nuts, Bolts and Bling
dougneiner
61
7.5k
Transcript
Reliability of Distributed Systems - Piyush Verma
Every product either dies a hero or lives long enough
to hit Reliability issues. 2
Customer Empathy No Chooran. Cost to Everything Architectures adapt to
$ Priority 01 02 03 04 3 Take it and Go
Service receives SMS User sends SMS Remind me to buy
milk at 6:30 PM to 53308 Cron gets Activated when time is right. Call the User 4 Sample Product
Service receives SMS User sends SMS Remind me to buy
milk at 6:30 PM to 53308 Cron gets Activated when time is right. Call the User 5 Sample Product: Inbound
Cron gets Activated when time is right. Call the User
6 Sample Product: Outbound Service receives SMS User sends SMS Remind me to buy milk at 6:30 PM to 53308
7 Inbound Connection
— Leslie Lamport https://www.microsoft.com/en-us/research/uploads/prod/2016 /12/Distribution.pdf “A distributed system is one
in which the failure of a computer you didn’t even know existed can render your own computer unusable” 8
Four Flavors of Failure 9 Disk Network CPU Memory
Network is Reliable Intra-LAN latency is ~ Zero Network is
Homogeneous Network cost is Zero 01 02 03 04 10
11 Scope of Failures: Again
At-least one server is online All servers are below 100%
All servers are responding within x ms. All of the above. 01 02 03 04 12
13 #1 Server is Unavailable
14 Replication Available
15 Replication Available
16 Available != Load Balanced
17 Load Balanced
18 Architecture of a Balancer
GCP AWS On-Prem Azure Who uses What? 19
Trilemma 20
Trilemma 21 Available Economical Endurable
22 Available + Load Balanced
Load Balancing 23
24 Monty Hall Problem: Was Marilyn vos Savant, right?
25 Server-side Load Balancing Example: Fabio
26 Look-aside Load Balancing Example: Consul/ DNS
27 Client-side Load Balancing Example: Ribbon
28 Client-side Load Balancing Example: Ribbon + Curator
29 Problems
30 Load Shedding
— Tyler McMullen https://www.infoq.com/presentations/load-balancing/ “Load Balancing is almost Impossible” 31
Alternate Reliability 32
33 Asynchronous Architectures
34 Asynchronous Architectures Example: RabbitMQ Kafka Kinesis SQS
Cron gets Activated when time is right. Call the User
35 Sample Product: Outbound Service receives SMS User sends SMS Remind me to buy milk at 6:30 PM to 53308 Part 1
36 Outbound
37 Scope of Failure: Outbound
38 Retries
39 Retries: Transient Failures
40 Exponential Backoff: Short term Transient Failures ✋ ✋ ✋
✋ ✋
41 Circuit Breaking: Long Term Transient Failures
42 Revisited
Dilemma 43 At-least Once Exactly-Once At-most Once
44 At-most once delivery
45 At-least once delivery
46 Exactly once delivery
47 Exactly once delivery = At-least-once Delivery + Exactly-once Processing
Keys to Only-Once delivery 48 Atomic Window Idempotent
Out-of-Order delivery 49
Revisit 50
Service receives SMS User sends SMS Remind me to buy
milk at 6:30 PM to 53308 Cron gets Activated when time is right. Call the User 51 Sample Product
52 Problems of State
53 Problems of State
54 Locked /Serialization
55 Master/ Master/ Slave
56 Clustering
Scalability 57 Data Replication Reduced Communication Logic/Data Decentralization
CAP Theorem [Sab topi pehna rahe] 58
59 Trilemma Available Partition Consistent
PACeLC Theorem 60
Dilemma 61 Consistency Latency
62 Revised Flow
63 What about Spanner? What about Calvin?
Reliable System 64 Scalable Correct Transparent
Access Transparency Location Transparency Concurrency Transparency Failure Transparency 01 02
03 04 65
Size Scalability Geographical Scalability 01 02 66
Summary 67 Consistent Available Economical Low Latency
68 All Put Together
Embrace your Bugs No Silver Bullet Cost to Everything Product
First 01 02 03 04 69
Does anyone have any questions?
[email protected]
Thanks 70