Slide 1

Slide 1 text

design for cloud mathias meyer, jax 2012

Slide 2

Slide 2 text

No content

Slide 3

Slide 3 text

Me — infrastructure — code — databases @roidrage www.paperplanes.de

Slide 4

Slide 4 text

No content

Slide 5

Slide 5 text

the cloud

Slide 6

Slide 6 text

No content

Slide 7

Slide 7 text

No content

Slide 8

Slide 8 text

No content

Slide 9

Slide 9 text

10,000 feet

Slide 10

Slide 10 text

amazon web services

Slide 11

Slide 11 text

ec2

Slide 12

Slide 12 text

on-demand computing

Slide 13

Slide 13 text

api

Slide 14

Slide 14 text

pay as you go

Slide 15

Slide 15 text

multiple regions

Slide 16

Slide 16 text

multiple datacenters

Slide 17

Slide 17 text

high cpu vs. high memory

Slide 18

Slide 18 text

elastic block store

Slide 19

Slide 19 text

more aws products s3 cloudfront cloudformation cloudwatch rds auto scaling simpledb route 53 load balancing queue service notification service elastic mapreduce

Slide 20

Slide 20 text

what’s scalarium?

Slide 21

Slide 21 text

automates: setup configuration one-click deploy

Slide 22

Slide 22 text

...for ec2 ...on ec2

Slide 23

Slide 23 text

the dream: configure a cluster push a button boom!

Slide 24

Slide 24 text

No content

Slide 25

Slide 25 text

No content

Slide 26

Slide 26 text

configuration + cookbooks/manifests + chef/puppet = configured cluster

Slide 27

Slide 27 text

configuration: chef server rightscale json scalarium

Slide 28

Slide 28 text

No content

Slide 29

Slide 29 text

in the beginning...

Slide 30

Slide 30 text

Scalarium

Slide 31

Slide 31 text

No content

Slide 32

Slide 32 text

No content

Slide 33

Slide 33 text

No content

Slide 34

Slide 34 text

ec2 is not a traditional datacenter

Slide 35

Slide 35 text

No content

Slide 36

Slide 36 text

No content

Slide 37

Slide 37 text

multi-tenant

Slide 38

Slide 38 text

high chance of failure

Slide 39

Slide 39 text

faulty instances

Slide 40

Slide 40 text

datacenter outage

Slide 41

Slide 41 text

network partition

Slide 42

Slide 42 text

more instances = higher chance of failure

Slide 43

Slide 43 text

mtbf

Slide 44

Slide 44 text

21/04/2011

Slide 45

Slide 45 text

No content

Slide 46

Slide 46 text

7/8/2011

Slide 47

Slide 47 text

No content

Slide 48

Slide 48 text

don’t fear failure

Slide 49

Slide 49 text

plan for failure

Slide 50

Slide 50 text

test failure

Slide 51

Slide 51 text

No content

Slide 52

Slide 52 text

plan for recovery

Slide 53

Slide 53 text

mttr

Slide 54

Slide 54 text

No content

Slide 55

Slide 55 text

disaster recovery plan

Slide 56

Slide 56 text

multi-datacenter deployments

Slide 57

Slide 57 text

replication

Slide 58

Slide 58 text

No content

Slide 59

Slide 59 text

multi-region deployments

Slide 60

Slide 60 text

$$$

Slide 61

Slide 61 text

relax consistency requirements

Slide 62

Slide 62 text

latency

Slide 63

Slide 63 text

keep data local

Slide 64

Slide 64 text

No content

Slide 65

Slide 65 text

keep data in memory

Slide 66

Slide 66 text

cache is king

Slide 67

Slide 67 text

use raids for ebs

Slide 68

Slide 68 text

use local storage

Slide 69

Slide 69 text

use bigger instances

Slide 70

Slide 70 text

scalarium is ~monolithic

Slide 71

Slide 71 text

cloud ≠ monolithic

Slide 72

Slide 72 text

cloud ☠ centralized

Slide 73

Slide 73 text

small, independent services

Slide 74

Slide 74 text

self-contained

Slide 75

Slide 75 text

independently deployable

Slide 76

Slide 76 text

well defined apis

Slide 77

Slide 77 text

http/json thrift protocol buffers

Slide 78

Slide 78 text

sinatra dropwizard spark scalatra

Slide 79

Slide 79 text

fail fast

Slide 80

Slide 80 text

retry

Slide 81

Slide 81 text

internal queues

Slide 82

Slide 82 text

design for resilience

Slide 83

Slide 83 text

“think about your software running.” theo schlossnagle, omniti

Slide 84

Slide 84 text

understand your code’s breaking points

Slide 85

Slide 85 text

isn’t all that what you do at large scale?

Slide 86

Slide 86 text

cloud == large scale

Slide 87

Slide 87 text

scalarium today

Slide 88

Slide 88 text

scalarium runs on scalarium

Slide 89

Slide 89 text

No content

Slide 90

Slide 90 text

No content

Slide 91

Slide 91 text

No content

Slide 92

Slide 92 text

lack of visibility

Slide 93

Slide 93 text

don’t fall for service level agreements

Slide 94

Slide 94 text

amazon only handles infrastructure

Slide 95

Slide 95 text

how you build on it is up to you

Slide 96

Slide 96 text

fun fact

Slide 97

Slide 97 text

amazon.com is served off ec2

Slide 98

Slide 98 text

thank you!