Professional Documents
Culture Documents
15.1
Cloud Computing
• Cloud computing – “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) that can be rapidly
provisioned and released with minimal management effort or
service provider interaction.”*
* “Above the Clouds: A Berkeley View of Cloud Computing,” University of California at Berkeley
Technical Report No. UCB/EECS-2009-28. 15.3
Difference between renting physical
servers remotely and cloud computing
• In cloud computing, you get a virtual machine running on
servers with your selected OS running on top of
virtualization software.
• There could be other users on servers.
• You access servers through a web service/web site.
• You pay for specific time used on processor, storage
devices and bandwidth/network.
• Cloud computing focuses on virtualization and service
orient approach and making it economical fro companies to
use a third party cloud provider to maintain hardware and
software on a on-demand basis.
15.4
Relationship to Grid computing
• Grid computing – using geographically distributed
computing resources collaboratively began as a concept in
the mid 1990’s with the growth of high speed networks and
the Internet.
• Began in the 1990s as a research concept to provide
collaborative computing
• The word “grid” came from the idea that grid computing
would provide computing power on demand through the
Internet in the same way as electrical power come from a
distributed electrical Grid utility.
• Cost of usage was not a driving force and usually no costs
charged.
15.5
Grid and Cloud Computing
• Both Grid computing and Cloud computing take advantage
of the Internet.
15.6
Utility computing resources
Utility computing suggested by John McCarthy in 1960s:
“computation may someday be organized as a public utility."
(Wikipedia).
Grid took it up idea in on-demand computing
Cloud computing followed through with:
1.Maturing of virtualization and service-oriented technologies
2.The growth of large underutilized data centers.
15.7
Technologies underpinning
Cloud computing
15.8
(Hardware) Virtualization
• Method hiding the physical characteristics of a
computer platform. User sees an abstract platform
• Hypervisor - Software that controls this
virtualization. (Word originally derived from 1960’s
“supervisor”.)
• Virtual machine (VM) - a “completely isolated guest
operating system installation within your normal
host operating system”
• User’s programs execute on this virtual machine
but has some access to underlying hardware as
controlled by hypervisor.
• Different OS’s can be provided to individual users
• Performance reduced (how much?) but provides
users with the illusion of their own platform. Users
isolated from each other.
http://en.wikipedia.org/wiki/Hardware_virtualization
http://en.wikipedia.org/wiki/Virtual_machine 15.9
Full and Platform Virtualization
• Full virtualization – complete simulation of underlying
hardware – all instructions, etc.
– Hardware-assisted virtualization – hardware architectural
support provided to allow virtualization
Server virtualization?
15.10
VMware
• A company started in
1998 providing
virtualization software,
notably hypervisors
• Offers a number of
products
– Cloud Foundry -- free,
open source cloud
computing platform as a
service (PaaS) software
http://en.wikipedia.org/wiki/VMware
http://www.vmware.com/ 15.11
Hypervisor Example: Xen hypervisor
and Xen cloud platform
Open source hypervisor
for x86, x86-64, Itanium,
Power PC, and ARM
processors.
Supports various OS’s
including Linux and
Windows
* http://en.wikipedia.org/wiki/Cloud_computing
15.14
Infrastructure as a Service (IaaS)
Advantages
• Access to preconfigured environment
• Use of latest technology
• Reduced cost and risk of having third party maintain resources
• No capital investment
• No IT personal to maintain remote hardware/software
• Able to manage peak demand as needed without having to
purchase a larger system that would be underutilized at other
times
• Secure – security handed by provider
Disadvantages ? Discuss
15.17
Salesforce.com
15.20
Amazon Web
Services
(AWS)
Amazon started as an
on-line bookstore
in1994/5
Large server farms for
their online business,
led to offering servers
to users through
Google moved into cloud
Amazon Web Services computing in same way having
(AWS) in 2006. large available server farms.
15.21
• Amazon led cloud deployment with their AWS
• They realized their large underutilized data centers
could be put to good use by providing cloud
computing to customers.
• AWS - a collection of remote computing (web)
services offered over the Internet (HTTP with
REST/SOAP protocols)
• Notable:
• Amazon EC2 – Amazon Elastic Compute Cloud – rent
virtual computers to run your own applications. Launched
2006. Full production in 2008.
• Amazon S3 – Amazon Simple Storage Service - provides
storage thro web service interfaces. Launched 2006
http://en.wikipedia.org/wiki/Amazon_Web_Services
http://en.wikipedia.org/wiki/Amazon_EC2 15.22
http://en.wikipedia.org/wiki/Amazon_S3
Amazon Elastic Compute Cloud (EC2)
• Uses Xen virtualization to create an instance
• Various packaged instances, see next
• Computing power defined by Elastic Compute Unit
(ECU)
– One EC2 Compute Unit equivalent CPU capacity of a
1.0-1.2 GHz 2007 Opteron or 2007 Xeon processor.
– 33.5 EC2 Compute Units = 2 x Intel Xeon X5570, quad-
core “Nehalem” architecture
• “Elastic” implies can quickly grow and shrink
available computing power (within minutes) – user
has to use AWS APIs and commands do do this?
15.23
AWS instances (2011)
http://aws.amazon.com/ec2/
• Standard Instances
– Small Instance (Default) 1.7 GB memory, 1 EC2 Compute Unit (1 virtual
core with 1 EC2 Compute Unit), 160 GB local instance storage, 32-bit
platform
– Large Instance 7.5 GB memory, 4 EC2 Compute Units (2 virtual cores
with 2 EC2 Compute Units each), 850 GB local instance storage, 64-bit
platform
– Extra Large Instance 15 GB memory, 8 EC2 Compute Units (4 virtual
cores with 2 EC2 Compute Units each), 1690 GB local instance storage,
64-bit platform
• Micro Instances – to add burst capacity
• High-Memory Instances – increased memory
• High-CPU Instances – increase CPU performance
• Cluster Compute Instances – cluster configurations
• Cluster GPU Instances - GPU cluster configurations 15.24
Amazon Simple Storage Service (S3)
Provides storage thro web service interfaces. Launched 2006
Data organization
– Write/read/delete objects (1 byte to 5 TB each)
– Each object stored in bucket retrieved by unique developer
assigned key
– Buckets stored in one of several regions: US Standard, EU
(Ireland), US West (Northern California), Asia Pacific
(Singapore), Asia Pacific (Tokyo)
– Objects kept in one region (unless you transfer them out)
– Authentications mechanism – private, public or rights to
specific user
http://aws.amazon.com/s3/ 15.25
Amazon Simple Storage Service (S3)
continued
http://aws.amazon.com/s3/ 15.26
S3 costs (2011)
Sliding Scale. Briefly:
Storage
First 1 TB / month $0.140 per GB $0.093 per GB
….
Over 5000 TB / month $0.055 per GB $0.037 per GB
Apart from
software for
Windows
platforms,
provides data
centers in US,
Europe and
Asia. 15.28
http://www.microsoft.com/windowsazure/features/
15.29
Eucalyptus
15.30
Some key Cloud Computing Issues
15.31
15.32
Using Cloud computing in Distributed
High Performance Computing (HPC)
AWS EC2 provides instances for HPC:
•High-CPU Instances
– High-CPU Medium Instance 1.7 GB of memory, 5 EC2 Compute Units (2 virtual
cores with 2.5 EC2 Compute Units each), 350 GB of local instance storage, 32-bit
platform
– High-CPU Extra Large Instance 7 GB of memory, 20 EC2 Compute Units (8 virtual
cores with 2.5 EC2 Compute Units each), 1690 GB of local instance storage, 64-bit
platform
•Cluster Compute Instances
– Cluster Compute Quadruple Extra Large 23 GB memory, 33.5 EC2 Compute Units,
1690 GB of local instance storage, 64-bit platform, 10 Gigabit Ethernet
•Cluster GPU Instances
– Cluster GPU Quadruple Extra Large 22 GB memory, 33.5 EC2 Compute Units, 2 x
NVIDIA Tesla “Fermi” M2050 GPUs, 1690 GB of local instance storage, 64-bit
platform, 10 Gigabit Ethernet
One EC2 Compute Unit (ECU) provides the equivalent CPU capacity of a
1.0-1.2 GHz 2007 Opteron or 2007 Xeon processor. 15.33
MapReduce
“Introduced by Google in 2004 “to support distribute computing
on large data sets on clusters of computers.”*
Computational processing occurs using two basic steps:
"Map step: The master node takes the input, partitions it up into smaller
sub-problems, and distributes those to worker nodes. A worker node may
do this again in turn, leading to a multi-level tree structure. The worker node
processes that smaller problem, and passes the answer back to its master
node.
"Reduce step: The master node then takes the answers to all the sub-
problems and combines them in some way to get the output – the answer to
the problem it was originally trying to solve.”*
Used at Google. Apache Hadoop is an implementation of MapReduce.
MapReduce using Hadoop available in AWS “Amazon Elastic MapReduce”
* http://en.wikipedia.org/wiki/Mapreduce
15.34
Questions?
15.35
Quiz Questions
What type of Cloud service is Amazon Web
services (AWS)?
15.36
What type of Cloud service is Microsoft Azure?
15.37
Reading materials
Cloud fundamentals:
15.39
HPC:
MapReduce/Hadoop …
15.40