Cloud Service Modelsand PerformanceAng Li09/13/2010

RoadmapRe-cap on cloud computing

Classes of cloud providers

Common services offered by cloud providers

Cloud performance comparison

Re-cap on cloud computingWhat is cloud computing?

◦Cloud application: software-as-a-service

◦Cloud provider: hardware and software infrastructure that supports the cloud applications

Benefits of cloud computing◦Pay-as-you-go◦On-demand scaling

Classes of cloud providers

Platform as a Service(PaaS)

Infrastructure as a Service (IaaS)

Infrastructure-as-a-Service providersOffer near bare-metal virtual

machines◦You can ssh, get root privilege, install

new software, do whatever you want◦Can use the APIs provided by the OS◦Highly flexible and customizable

Charge by machine-hour◦ (How many machines you use) X

(how long you use them)◦You can afford it!

Demo time!Amazon AWS

Platform-as-a-Service providersOffer a sandbox

environment◦Upload a program, they run it

for you◦Can only use the APIs provided

by the environment

Charged by CPU utilization◦Pay only how much resource

you use◦A few free hours per day!

Demo time!AppEngine

Windows Azure: a combination of bothOffer a sand-box environment of

C# (PaaS)Charge by machine-hour (IaaS)

Demo time!

DiscussionWhat are the pros and cons of

IaaS and PaaS?

Which one do you prefer?◦Your homepage◦E-business application◦Video processing

Services offered by a cloudElastic compute cluster

Persistent storage

Intra-cloud and wide-area network

and others…◦MapReduce service◦CDN service

Elastic compute clusterWhere your application is running

◦VM or sandbox environmentWhy cluster?

◦Multiple “instances” can be running your application simultaneously

Why elastic?◦You can add new instances or

remove existing ones with very short latency

Scaling the compute clusterOpaque scaling

◦User can manually increase/decrease the number of instances

◦Alternatively, she can set up a scaling policy

◦IaaS providers (including Azure)

Transparent scaling◦Scaling happens automatically

(magically)◦PaaS providers

Persistent storageWhere to store your application’s

dataWhy persistent?

◦Local storage (VM disk) is not reliableDifferent types of storage service

◦Table, blob, queue, etc.◦Highly scalable and available

Access the storage services

Storage APIs◦HTTP-based

GET https://sdb.amazonaws.com/?Action=PutAttributes&Attribute.1.Name=data&Attribute.1.Value=haha&AWSAccessKeyId=[valid access key id]&DomainName=table&ItemName=k…

◦Library callDatastoreService datastore = DatastoreServiceFactory.getDatastoreService();

Key k = KeyFactory.createKey(table, key);Entity e = new Entity(k);e.setProperty("data", “haha");datastore.put(e);

Networking serviceIntra-cloud network: between two

instances or between an instance and the storage service

Wide-area network: between a cloud instance and the end user

Intra-cloud networkWithin a data center

◦High bandwidth, low latency

Wide-area networkEvery provider has multiple locations to host

cloud applications

Application content can be served from the closest location◦ Why is this a good thing?

Re-cap: cloud services

Intra-cloud network

Storage service

Computation service

Wide-area network

Web application

CloudCmp: comparing cloud providersMotivation

◦Provide shopping advice Analogy: should I buy an IBM or a Dell or

a Mac?

◦Identify performance problems

Challenges◦What to compare?

Each provider is unique in some way…

◦How do we measure?

MethodologyIdentify the common services

◦We just did thisPick a few metrics for each

service relevant to application performance

Develop benchmarking task for each metric

Run the tasks on different providers and compare

Choose the providers to compareAWS, Rackspace, Azure, and

AppEngine

Metrics: elastic compute clusterBenchmark finishing time

◦Java-based benchmark tasksCost per benchmark

◦From per-hour price and billing APIScaling latency

◦Periodically allocate new instances

Metrics: persistent storageOperation latency

◦Client mattersCost per operationTime to consistency

◦Read-after-write test

Metrics: networking serviceIntra-cloud network

◦Bandwidth (iperf)◦Latency (ping)

Wide-area network◦Use PlanetLab nodes to simulate a

diverse user base◦Let each node ping each data center

of a cloud provider◦Optimal wide-area latency

Result: computation performance

What can we learn from the figure?◦ How do the performance differ across different clouds?◦ Within a provider, how do performance differ across different

instance types? Is this enough information to choose provider/instance?

◦ How about cost?

Result: computation cost

What can we learn from this figure?◦Is some provider particularly cost-

effective?◦Which instance type should I choose

within a provider?

Result: scaling efficiency

How is the scaling latency (high/low)?

Linux vs Windows?

Result: storage (table)

The storage services show high variation◦Median – 30ms, 90 percentile – 60ms

Is it good or bad? Why?

Result: intra-cloud network

Intra-cloud bandwidth varies across different providers◦From 200Mbps to 800Mbps

What might be the cause?

Result: wide-area network

Wide-area network latency also varies a lot◦C3 is much better than the others◦What might be the reason?

Result summaryNo provider stands out

◦C1 has the highest network bandwidth, while its instance is not the most cost-effective

◦C2 has the most powerful instances, while its network bandwidth is low

◦C3 has the lowest wide-area network latency, while its storage is slower than others

It is not trivial to shop for a cloud provider!◦Many research challenges in developing a

sound mechanism to select the best provider for an application

SummaryTwo main classes of cloud providers

◦IaaS: bare-metal virtual machines◦PaaS: sand-box environment

Four common services◦Elastic compute cluster◦Persistent storage◦Intra-cloud network◦Wide-area network

No provider has the best performance over all services