fabien hermenierfhermeni.github.io/sacc/sacc-introduction.pdf~5 racks go wonky (40-80 machines see...
TRANSCRIPT
Cloud compting101
Fabien Hermenierimage credit http://eyepluscamera.files.wordpress.com/
1
was cloud computing needed ?
2
Mainframes
3
Then came with affordable PCs
Then we spread out the load for security, performance, manageability
Then we bought tons of servers to support load spikes
4
5
Amazon X-mas 2013 426 items sold each second6
Where is energy spent ?
7
episode 0 rise of the cloud8
Cloud computing is a model for enabling ubiquitous, convenient, on-demand network access to a shared pool of configurable computing resources that can be rapidly provisioned and released with minimal management effort or service provider interaction.
2011
“”9
1 self-provisioning, no human intervention
On-demand self-services
10
11
2availability over the network
standard mechanisms
broad network access
12
3multi-tenant
virtual or physical resources on-demand allocation
location independance
resource pooling
13
reserved instances (yearly based) on-demand instances (hourly based)
hotspot instances (market based) 14
Amazon EC2 HotSpot instances
bid over the market price to get the instance15
4fast (de-)allocation of resources scale to infinity
rapid elasticity
16
verti
cal e
lastic
ity
Tiers 1
Tiers 2
Tiers 3
17
horizontal elasticity
Tiers 1
Tiers 2
Tiers 3
18
5 metering capabilities transparent reporting
measured service
19
20
and I will call it cloud computing
21
Cloud Computingorigins
22
If computers of the kind I have advocated become the computers of the future, then computing may someday be organized as a public utility just as the telephone system is a public utility... The computer utility could become the basis of a new and important industry.
John McCarthy, 1961
“”
23
cluster computingloosely coupled co-located servers
single tenant non-interactive workload
rigid jobs80s
24
Clo
ud o
r no
t ?
on demand self-services
broad network access
resource pooling
rapid elasticity
measured service
25
Clo
ud o
r no
t ?
on demand self-services
broad network access
resource pooling
rapid elasticity
measured service
26
grid computingIan Foster et al. 2001
27
Pow
er G
rid
Ana
logy
28
virtual organisation heterogeneous hw. multiple applications abstract resources
doing * at
Power grid Computing grid
multiple providers heterogeneous sources
multiple clients abstract source
large scaleindependencelocation
live consumption batch jobs
29
Throughput(MB/s)
TransferThroughput2014-10-2212:40to2014-10-2312:40UTC
alice atlas cms lhcb
13:00
14:00
15:00
16:00
17:00
18:00
19:00
20:00
21:00
22:00
23:00
00:00
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
09:00
10:00
11:00
12:00
0k
5k
10k
15k
20k
25k
Worldwide LHC Computing grid
170 centres to analyse 30 PB / year30
Clo
ud o
r no
t ?
on demand self-services
broad network access
resource pooling
rapid elasticity
measured service
31
Clo
ud o
r no
t ?
on demand self-services
broad network access
resource pooling
rapid elasticity
measured service
32
Application Service Provider
service oriented
pay as you go
95+
1Client
2Client
3Client
remote access to dedicated applications
33
Clo
ud o
r no
t ?
on demand self-services
broad network access
resource pooling
rapid elasticity
measured service
34
Clo
ud o
r no
t ?
on demand self-services
broad network access
resource pooling (not real hw resources)
rapid elasticity
measured service
35
computers on demand.2002Deploy full custom stacks (OS to applications)
36
to (re)deploy reproducible network experiments
multi-tenant, (limited on purpose) resource pooling,
37
to (re)deploy reproducible network experiments
100Mb/s10ms
10ms50ms,
5% loss
38
to (re)deploy reproducible network experiments
10ms
10ms
50ms, 5% loss
39
Clo
ud o
r no
t ?
on demand self-services
broad network access
resource pooling
rapid elasticity
measured service
40
Clo
ud o
r no
t ?
on demand self-services
broad network access
resource pooling
rapid elasticity
measured service
41
S O Aervice
orientedrchitecture
2001+
composable unassociated, loosely coupled units
42
exponential grows since 2001 private and public services to support its growth
43
Two pizza rule
If a team can’t be fed by two pizzas then it is to big
- Jeff Bezos (founder/ CEO of amazon.com)
44
800 x
tons of API, mini-services devoted to automation, flexibility, on-demand services for public and private use
45
2006
scalable web services for other websites or client-side applications
46
SOAP & REST over HTTP pay as you go elastic *-oriented services
*data, network or computation47
Clo
ud !
on demand self-services
broad network access
resource pooling
rapid elasticity
measured service
48
RECAP49
I have a dream, it was about Utility Computing “ ”John McCarthy - 1961
50
web + grid computing + resources on demand + service oriented architectures
cloud computing (2006)
51
can we talk about cloud computing now ?
52
?aaS53
web access to commercial sw. “one to many” model
customers don’t handle upgrades API for integration
Service as a Service
Saa
S
54
55
56
jailed runtime available to host applications generic or provider-specific APIs no control over the environment
Platform as a Service
Paa
S
57
58
$ heroku login…
$ git clone https://github.com/heroku/java-getting-started.git$ cd java-getting-started
$ heroku createCreating warm-eyrie-9006... done, stack is cedar-14http://warm-eyrie-9006.herokuapp.com/ | [email protected]:warm-eyrie-9006.gitGit remote heroku added
$ git push heroku master… http://warm-eyrie-9006.herokuapp.com/ deployed to Heroku
$ heroku ps:scale web=1
59
Infrastructure as a Service
low-level resources to deploy arbitrary software stacks complete control over its network, storage and OS
IaaS
60
61
62
63
Things will crash. Deal with it!
Assume you could start with super reliable servers (MTBF of 30 years) Build computing system with 10 thousand of those Watch one fail per day
Dean Keynote, LADIS 2009
“
”
64
~0.5 overheating (power down most machines in <5 mins, ~1-2 days to recover) ~1 PDU failure (~500-1000 machines suddenly disappear, ~6 hours to come back) ~1 rack-move (plenty of warning, ~500-1000 machines powered down, ~6 hours) ~1 network rewiring (rolling ~5% of machines down over 2-day span) ~20 rack failures (40-80 machines instantly disappear, 1-6 hours to get back) ~5 racks go wonky (40-80 machines see 50% packetloss) ~8 network maintenances (4 might cause ~30-minute random connectivity losses) ~12 router reloads (takes out DNS and external vips for a couple minutes) ~3 router failures (have to immediately pull traffic for an hour) ~dozens of minor 30-second blips for dns ~1000 individual machine failures ~thousands of hard drive failures slow disks, bad memory, misconfigured machines, flaky machines, etc. Long distance links: wild dogs, sharks, dead horses, drunken hunters, etc.
Typical first year for a new google cluster
Leslie Lamport
A distributed system is one in which the failure of a computer you didn't even know existed can render your own computer unusable
65
“”
66
Building fault tolerant services
deal with failures deal with inconsistency
be pessimistic
at every level
applications
runtimes
integration/security
database
servers
virtualisation
server HW
storage
network
you
man
age ol’
school IT
67
SaaS
applications
runtimes
integration/security
database
servers
virtualisation
server HW
storage
network
man
aged
by ve
ndor
68
PaaS
applications
runtimes
integration/security
database
servers
virtualisation
server HW
storage
network
you manage
man
aged
by ve
ndor
69
Open-source PaaS stacks
IaaS
applications
runtimes
integration/security
database
servers
virtualisation
server HW
storage
network
you
man
age
man
aged
by ve
ndor
71
Open-source IaaS stacks
(2008+)
cloudstack
(2008+)
(2010+) (2012+)
72
73
vendor lock-in
IaaS PaaS SaaS
Deployment models
74
public cloudgeneral availability to everyone
the “real” cloud reduced costs trust issues ?
75
76
cloud computing vs.
fog of war
77
Trust in megive me your code & data
78
I’m aware read my mails
what is my is hacked ?
private cloudself hosted cloudworldcompany SA
might reduce TCO stronger trust
better manageability
79
worldcompany SA hybrid cloud
80
multi-cloudsyou spread your application
avoid Single Point of Failures* take the benefits of each cloud
LB
81
inter-cloudsthey outsource your components
agreements between the providers“cloud of clouds”
82
fog computing
83
community cloudprivate cloud by and for
multiple organizations
84
RECAP85
CLOUD IS ABOUT REDUCING COSTS
86
CLOUD IS ABOUT SCALABILITY
87
88
CLOUD IS ABOUT RESILIENCY
CLOUD IS ABOUT TRUST