Understanding Modern Web Service Deployment in EC2 and Azure Keqiang He Alexis Fisher Liang Wang Aaron Gember Aditya Akella Thomas Ristenpart University of WisconsinMadison ID: 493493
Download Presentation The PPT/PDF document "Next Stop, the Cloud:" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Next Stop, the Cloud: Understanding Modern Web Service Deployment in EC2 and Azure
Keqiang He, Alexis Fisher, Liang Wang, Aaron Gember, Aditya Akella, Thomas RistenpartUniversity of Wisconsin-Madison
1Slide2
Private
Datacenters
Co-location
Centers
Move to
IaaS
clouds
2Slide3
Incomplete view of cloud use3
Want a global, in-depth understanding
of
IaaS
cloud usage patternsSlide4
Fundamental questions
Who is using public IaaS clouds?Traffic patterns; network design/traffic engineeringHow are these services using the cloud? Impact of failures; ways to improve availability Design of new systems/servicesHow can quality of experience be improved? Deployment design (e.g., ideal region/zone usage)
4Slide5
First in-depth empirical study examining
IaaS cloud usage patterns and identifying ways tenants could better leverage IaaS clouds5Slide6
DNS
Server
CDN
VM
IaaS
cloud environments
6
VM
LB
Proxy
PaaS
Node
VM
Platform-as-a-Service (
PaaS
)Slide7
Datasets
University packet captureDeep, but possibly atypical, perspectiveAlexa subdomains DNS recordsBroad, high-level perspective7
Top 1
million
Subdomains
Zone transfer
DNS
records
1.4TB Capture
IP Ranges
Conn
records
Campus
TrafficSlide8
Fundamental questions
Who is using public IaaS clouds?Traffic patterns; network design/traffic engineeringHow are these services using the cloud? Impact of failures; ways to improve availability Design of new systems/servicesHow can quality of experience be improved? Deployment design (e.g., ideal region/zone usage)
8Slide9
Cloud-using domains
Non-cloud-using domainsHow many popular web services use the cloud?
9
9.
13.
29.
35.
36.
2 of 68
3 of 142
4 of 181
18 of 24
14 of 89
43% in top
250K domains
Cloud availability and performance
issues may impact over 4%
of popular web servicesSlide10
Traffic Volume (Bytes)
Which (locally) popular
web services use the cloud?
10
Cloud availability and performance
affects a substantial fraction
of popular web services
68.2%
1.7%
1.1%
0.7%
0.6%
atdmt.com
s-msn.com
msecnd.net
3.1%
2.4%
2.3%
1.5%
1.4%Slide11
What type of content is served by(locally) popular web services?
11
text/html
text/plain
image/jpeg
app/flash
app/octet-stream
24.1%
23.4%
10.6%
8.7%
7.8%
HTTP Content Type
Top 100 cloud-using domains account for
≈80% ≈100%
of HTTP flows
Compression could be employed
to save WAN bandwidth
and improve delivery latencySlide12
Fundamental questions
Who is using public IaaS clouds?Traffic patterns; network design/traffic engineeringHow are these services using the cloud? Impact of failures; ways to improve availability Design of new systems/servicesHow can quality of experience be improved? Deployment design (e.g., ideal region/zone usage)
12Slide13
DNS
Server
CDN
VM
IaaS
cloud environments
13
VM
LB
Proxy
PaaS
Node
VM
PaaSSlide14
VM
?
VM
VM
LB
Proxy
?
Which front ends are used?
14
sub.abc.com
A
72.44.32.1
sub.abc.com
CNAME
abc.elb.aws.com
abc.elb.aws.com
A
72.44.42.1
sub.abc.com
CNAME
proxy.heroku.com
proxy.heroku.com
A
72.44.62.1
?
PaaS
Node
DNS
Server
VM
LB
Proxy
PaaS
Node
String matching
on CNAME
DNS
recordsSlide15
Which front ends are used?
≈60% of ELB-using
subdomains
resolve to 2+ ELB IPs
≈4% of ELB IPs are shared
by 10+
subdomains
15
There is limited uptake
of value-added featuresSlide16
Which front ends are used?
169. 13.29.35.36.
VM
LB
LB
PaaS
VM
LB
PaaS
CDN
Failures of value-added features
will impact the most popular
EC2-using
subdomainsSlide17
Which locations are used?
17
DNS
records
Match IP against
region rangesSlide18
How many and which regions are used?
18Slide19
How many and which regions are used?
19Slide20
How many and which regions are used?
9.
13.
29.
35.
36.
1
2
1
1
2
7.
18.
20.
31.
3
5
1
5
Single region failures can
take down a large fraction
of cloud-using
subdomainsSlide21
How many and which availability zones are used?
Latency measurementsLow latency => instances are in same zoneHigh coverage, low accuracy (noisy)IP address correlationIPs are in the same /16 subnet => instances are in the same zoneHigher accuracy, low coverage21Slide22
How many and which availability zones are used?
22
9.
13.
29.
35.
36.
4
5
1
3
5
Single zone failures can take
down 16% of
subdomains
and cripple many othersSlide23
Fundamental questions
Who is using public IaaS clouds?Traffic patterns; network design/traffic engineeringHow are these services using the cloud? Impact of failures; ways to improve availability Design of new systems/servicesHow can quality of experience be improved? Deployment design (e.g., ideal region/zone usage)
23
Many popular web services are not deployed optimally for their usersSlide24
Future measurementsUse
of other clouds (e.g., Rackspace)Use of other EC2 and Azure featuresBack end deploymentsPacket captures from other vantage points24Slide25
SummarySystematic method for studying IaaS
cloud use4% of popular web services use EC2/AzureVMs are the most popular EC2 front end, but value-added features are used by top domainsLimited region and zone use (and diversity) makes web services vulnerable to failures25Slide26
Are web services deployed near users?
Compare a domain’s customer country (according to Alexa) and region usage47% of subdomains are not deployed in the same country as users32% subdomains are not deployed on the same continent as users26Slide27
27
What is the user perceived network performance for different regions/zones?
Regions should be selected
based on users’ locations