COMP5349– Cloud Computing
Week 1: Introduction to Cloud Computing
Dr. Ying Zhou School of Computer Science
Outline
Cloud in layman term A brief history of Cloud
Cloud official definition The Role of Data Centers
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-2
What is “Cloud”?
Informally, we may view cloud computing as a way of renting IT resources
Through Internet/Web
Has an innovative way to specify, measure and charge the rented
resources
Many other features…
Not every kind of IT resources renting is called cloud Lease from Dell to equip our labs
Rent some space from your ISP to set up a website
Some forms of renting are closely related to Cloud and are considered as cloud’s competitor
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-3
What are IT resources?
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-4
To run a renting business
A way to package/measure/quantify your rental product/service
Leasing a Dell Desktop with certain specifications, each with different price
tag
A way to charge the customer
Hourly/daily/monthly/yearly rate Subscription
A way to deliver the produce/service Truck, courier, pickup, or Internet
A way to guarantee your product/service meet the client’s requirements
There are other forms of IT resources renting
The particular form ”Cloud” comes after supporting technologies are mature, and of course, good incentives for providers and market needs.
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-5
XML based SOAP web services was proposed in 1998
Amazon published a few SOAP bases services in 2002
AWS launched three services in 2006: EC2, S3 and SQS
Salesforce Now Live on Amazon Web Services Cloud Infrastructure in Australia (Oct. 2017)
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-6
Other major players
Microsoft
Microsoft Azure is officially released in 2010
It is now the second largest cloud provider with a market share less than half of the market share of AWS
Google
Google cloud platform
Google App Engine was released in 2008 (an early PaaS service) Google Cloud Platform was launched in 2011
IBM
The latest one is called Bluemix and was released in 2014 IBM cloud has been through many other versions
IBM cloud is not that well recognized in general public
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-7
Outline
Cloud in layman term
Cloud official definition
Various model
Specification, enabling technologies and pricing Incentives from provider and consumer perspectives
The Role of Data Centers
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-8
Cloud Computing– a Broad Definition
A definition by US Governments’ National Institute of Standard and Technology
“Cloud computing is a model for enabling ubiquitous, convenient, on- demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services)”
In early days, we tend to differentiate three different models Infrastructure as a Service (IaaS)
Platform as a Service (PaaS)
Software as a Service (SaaS)
There are many new services and
Many providers are not restricted by a single service model Many services cannot be categorized easily
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-9
A Service Provider’s offer: Azure
https://docs.microsoft.com/en-us/azure/architecture/aws-professional/services
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-10
SaaS Examples
Software as a Service (SaaS): The consumer uses an application, but does not control the operating system, hardware or network infrastructure on which it’s running.
Applications are restricted to business applications or applications that may normally installed in a business network or personal computer
Examples
• Business applications: CRM solutions from salesforce.com
• Business/Personal applications: Gmail, Google Doc, etc.
SaaS in many ways are different to the others and many times are not included when we say cloud
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-11
PaaS Examples
Platform as a Service (PaaS): The consumer uses a hosting environment for their applications. The consumer controls the applications that run in the environment (and possibly has some control over the hosting environment), but does not control the operating system, hardware or network infrastructure on which they are running. The platform is typically an application framework.
AWS Elastic MapReduce
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-12
IaaS Examples
› Infrastructure as a Service (IaaS): The consumer uses “fundamental computing resources” such as processing power, storage, networking components or middleware. The consumer can control the operating system, storage, deployed applications and possibly networking components such as firewalls and load balancers, but not the cloud infrastructure beneath them.
…
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-13
Spectrum of Cloud Services
Lower-level,
Less management
Higher-level, More management
Azure
Utility computing
From Berkeley Cloud presentation: http://berkeleyclouds.blogspot.com/
EC2
AppEngineForce.com GoogleApps
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-14
SaaS service specification and pricing
SaaS
The service specification depends on the actual application, it could be the number of user account supported, the size of storage, etc
The pricing is usually subscription based, e.g. monthly or yearly price
https://products.office.com/en-au/compare-all-microsoft- office-products?tab=2 accessed 07/03/2018
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-15
IaaS Specification and Pricing
IaaS
The specification is similar to the general spec when you purchase a
computer. These include cpu speed, number of cores, memory, etc
At the beginning, most providers use fine grained pay-as-you-go hourly rate
Now many providers have even finer grained “Per Second Billing”[ https://aws.amazon.com/ec2/pricing/ accessed 07/03/2018]
https://aws.amazon.com/ec2/pricing/on-demand/ https://aws.amazon.com/ec2/instance-types/ http://www.zones.com/site/product/index.html?id=105374001
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) accessed 07/03/2018 01-16
Specification and Pricing
PaaS
Somewhere in between, could be fine grained hourly rate or
subscription based.
E.g. If you start a MapReduce cluster in Azure or AWS, you can specify how many node you want to have and the node type, you will be charged hourly (or secondly) based on those instances’ price.
https://cloud.google.com/appengin e/pricing
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-17
Service packaging technologies
IaaS
IaaS provides virtual machine together with storage and network as package
All providers use virtualization technology, the actual software used could be different
PaaS
Virtualization technology can be used if the platform is presented as VM +
some preinstalled software
Container technology may be used and the launching time could be greatly reduced
Provider may design their own software to let clients share an underlying platform
SaaS
Again virtualization technology maybe used to provide each customer one or
many VMs with preinstalled apps
Most providers write multi-tenancy based system to allow better resource utilization
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-18
Other services
Other services have their own way of describing, charging and enabling technologies
E.g. most storage services rely on company’s own implementation of a planet scale storage system: Azure storage, DynamoDB, etc
Storage charging is more complicated as it has both the static and dynamic part
Actual storage size
Number of queries
Consistency and other quality requirement
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-19
Service Delivery
All those XaaS models are delivered through Internet, with a web interface
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-20
Clouds Servicing Models
Private
(On-Premise) Applications
Runtimes
Security & Integration
Databases Servers Virtualization
Server HW Storage Networking
Infrastructure
(as a Service) Applications
Runtimes
Security & Integration
Databases Servers Virtualization
Server HW Storage Networking
Platform
(as a Service) Applications
Runtimes
Security & Integration
Databases Servers Virtualization
Server HW Storage Networking
Software
(as a Service) Applications
Runtimes
Security & Integration
Databases Servers Virtualization
Server HW Storage Networking
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-21
You manage
You manage
You manage
Managed by vendor(s)
Managed by vendor(s)
Managed by vendor(s)
Using the Renting Terms
All cloud serving models are based on the subject of renting If I rent you the whole computer (the virtual version) with Operating
System preinstalled, I am providing Infrastructure as a service
If I rend you some development environment and let you build your own application on top it, I am providing Platform as a service
If I rent you the whole application for you to use, I am providing Software as a service
How do you determining what you want?
Or, why, as a consumer, I should consider any such option? Or, why, as a provider, I should provide such services?
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-22
Incentives for Cloud Providers
IT resource utilization is usually low
Server’s cpu, memory, IO, networking are in idle state most of the
time
If each person only uses one quarter of their server’s computing power, why not provide a way to let four person sharing one computer?
Economy of Scale
Most cloud providers are companies with large
amount of IT facilities
Resource Cost in Cost in Ratio Medium DC Very Large DC
Network
$95 / Mbps / month
$13 / Mbps / month
7.1x
Storage
$2.20 / GB / month
$0.40 / GB / month
5.7x
Administration
≈140 servers/admin
>1000 servers/admin
7.1x
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
03-23
Incentives for Cloud Users
Cloud user
Better provisioning through elasticity and pay-as-you-go and
other fine-grained pricing models
Lift the burden of operational management
Installing certain software, e.g. a Hadoop cluster or GPU based tensorflow is time consuming and error prone
Upgrading existing software adds further headache CapEx vs. OpEx tradeoff
Example
Netflix: world’s leading Internet subscription service for movies and TV
shows
Netflix migrated from its own data centers to AWS in 2010
Capacity growth rate is accelerating, unpredictable
• Year on year customer growth is 52%, year on year customers using streaming is up
145% (from ~4M to ~11M).
Product lunch spikes– iPhone, WII, PS3, XBox
Datacenter is large inflexible capital commitment COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-24
The Provisioning Problem
It is very hard to predict usage and to provision sufficient capacities
Capacity
Demand
Time Time Unused resources
From Berkeley Cloud presentation: http://berkeleyclouds.blogspot.com/
Capacity
Demand
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-25
Resources
Resources
Elasticity and Pay-As-You-Go
Elasticity
The cloud allows scaling up and scaling down of resource usage on an ‘as-needed’ basis. Elapsed time to increase or decrease usage is measured in seconds or minutes
Pay-As-You-Go
Consumer is charged based on resourced they used (per instance or per cpu time), the charging unit has changed to hourly to secondly (actually minutely for most providers)
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-26
The famous Animoto Example
“They had 25,000 members on Monday, 50,000 on Tuesday, and 250,000 on Thursday. Their EC2 usage grew as well.
For the last month or so they had been using between 50 and 100 instances. On Tuesday their usage peaked at around 400, Wednesday it was 900, and then 3400 instances as of Friday morning.”
http://aws.typepad.com/aws/2008/04/animoto–scali.html
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-27
Netflix Auto-scaling Observations
request traffic over two days.
Corresponding number of servers over same two days.
The Netflix Tech Blog: Auto Scaling in Amazon cloud (Jan. 18, 2012)
http://techblog.netflix.com/search/label/autoscaling
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-28
Netflix Auto-scaling Observations (cont)
Aggregated CPU utilization during this time period:
Note that under load the aggregate CPU is essentially flat
The Netflix Tech Blog: Auto Scaling in Amazon cloud (Jan. 18, 2012)
http://techblog.netflix.com/search/label/autoscaling
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-29
Outline
Cloud in layman term
Cloud official definition The Role of Data Centers
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-30
Cloud is very physical
All rentable IT resources reside in data center What is a data center?
“A building or portion of a building whose primary purpose is to house a computer room and its support areas” [TIA 942]
But a computer lab is not a data center, your garage with several machines is not a data center, even our server room is not a data center
Large companies like Amazon, Google, Microsoft operated their daily business in data centers before cloud era.
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-31
Data center Fundamentals
Typical components of a data center
IT Equipment
Server, Switches, Storage, etc.
The way to organize and manage those equipment change all the time
Facility Equipment Power and cooling
Ancillary Systems
Access control, CCTV, fire alarm, Data Center Infrastructure
Management Systems
Depends on complexity and robustness of data centers, they are categorized into four different tiers
Tier 1 being the most basic one Tier 4 being the most robust one
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-32
Data center main components
Figure 1.1 in Data Center Virtualization Fundamentals
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-33
The IT component
http://www.morganclaypool.com/doi/abs/10.2200/S00516ED2V01Y201306CAC024
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-34
Typical Server Unit
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-35
Performance Consideration
Traditionally: Computer Systems optimized for Performance (e.g. throughput)
Nowadays, power is an increasingly important measure too Costs of data center (DC) are dominated by power and cooling
infrastructure
Carbon trading schemes will affect this even more Hence DCs to optimize for work done/Watt
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-36
How is DC Energy Efficiency Measured?
Commonly used metric for assessing data centers (DCs):
Power Usage Efficiency (PUE)
Total Facility Power IT Equipment Power
The average data center in the US has a PUE of 2.0 (Source: EPA – U.S. Environmental Protection Agency)
According to James Hamilton, many have even PUE > 3.0
High scale cloud services in the 1.2 to 1.5 range
PUE =
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-37
Data Center Organization
Figure 1.3 in Data Center Virtualization Fundamentals
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-38
Interdependent decision making
The number of servers per cabinet depends on the power distribution design.
A network design must be based on the knowledge of how many servers will the installed in each rack and how many interfaces they have.
The network devices’ physical position in the data center influences the structure’s cabling project.
Cabling can be laid out under the raised floor.
The raised floor can have a direct influence over the cooling system, which is usually the highest contributor to the power system.
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-39
Technology life cycle
Design decision depends highly on the life cycle of relevant component to ensure future evolution
Building: 10 to 15 years
Cabling: 7 to 10 years
Network: 3 to 5 years
Storage: 1 to 2 years
Server: 6 to 18 months
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-40
Data Center Authorities
Uptime Institute
The Telecommunications Industry Association (TIA)
Uptime Institute Data Center Industry Survey results 2016
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-41
Data Center and Cloud
Most cloud providers have their own data centers globally
Some providers may use data centers owned by others through
Colocation provider, e.g. Equinox is a colo provider many cloud providers such as Azure, IBM and AWS rent space from
One type of cloud provider may user another provider’s resources, e.g. Salesforce uses AWS in Australia
“Colo is a type of data centre where equipment, space, and bandwidth are available for rental to retail customers.”
–[https://en.wikipedia.org/wiki/Colocation_centre]
Here equipment mostly refer to power, cooling and security ones
The actual IT equipment belongs to the company who rented the space, e.g. an entire floor
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-42
Amazon’s global infrstructure
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-43
Company or Organization’s IT asset choices
Enterprise owned data center Private cloud
Colo-data center Private cloud
Cloud
Private hosted cloud, e.g. Amazon VPC (Virtual Privte
Cloud[https://aws.amazon.com/vpc/pricing/])
“Hybrid infrastructure models now the norm”(Uptime Institute Data Center Industry Survey results 2016)
Our university is an example of using hybrid infrastructure model We are big users of Microsoft and AWS
We also use colo providers
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-44
Cloud
History and current status
Various cloud models
Description of the services
How to package the services Pricing of the services
Data center as the physical part of cloud
Summary
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-45
Next Week: Homework Homework for week 1
We prepared a self test homework to help you assess your Python programming skills
You may consider withdrawing from this unit if you have difficulty in completing it independently within a couple of hours.
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-46
Next Week: Readings
Gustavo A., A. Santana: Data Center Virtualization Fundamentals: Understanding Techniques and Designs for Highly Efficient Data Centers with Cisco Nexus, UCS, MDS, and Beyond
Chapter 1, chapter 13
Matthew Portnoy: Virtualization Essentials, Second Edition
Jim Smith (Author), Ravi Nair , Virtual Machines: Versatile Platforms for Systems and Processes, Morgan Kaufmann; 1 edition (June 17, 2005)
Chapter 1
Chapter 8 System Virtual Machines
Xen and the Art of Virtualization. Paul Barham, Boris Dragovic, Keir Fraser, Steven Hand, Tim Harris, Alex Ho, Rolf Neugebauer, Ian Pratt, and Andrew Warfield. In Proceedings of the Nineteenth ACM Symposium on Operating systems principles (SOSP ’03), 2003.
COMP5349 “Cloud Computing” – 2020 (Ying Zhou)
01-47
Reference
Armbrustet al: Above the Clouds: A Berkeley View of Cloud Computing, TR EECS-2009-28, UC Berkeley, 2009.
Gustavo A., A. Santana: Data Center Virtualization Fundamentals: Understanding Techniques and Designs for Highly Efficient Data Centers with Cisco Nexus, UCS, MDS, and Beyond
Chapter 1
COMP5349 “Cloud Computing” – 2020 (Ying Zhou) 01-48