Evolution - not revolution

basheddockDéveloppement de logiciels

21 févr. 2014 (il y a 3 années et 10 mois)

71 vue(s)

Evolution
-

not revolution

Server resources are shared globally
instead of locally



Excess capacity for peak usage
can be shared



Backup, security and other
common functions can be shared



Failover is part of the underlying
infrastructure


Integration of existing technologies

Virtualization, Internet, databases,
etc. have all been around for many
years


Many companies have already
outsourced their IT infrastructure

Much of the Cloud can be seen as
just another form of outsourcing


issues such as SLA, uptime,
security are similar or identical


Broadband is the enabler that has made the Cloud possible










































By location




Public


Infrastructure is hosted by Cloud vendor and shared with other tenants


Lowest cost and lowest security




Private

Infrastructure is not shared; may be on
-
premises or hosted externally

Increased cost and increased security




Hybrid

Both Private and Public Cloud services used together, perhaps to retain critical
data on
-
premises

Cloud Bursting


Private Cloud with Public Cloud to cover peak usage (not to be
confused with IBM’s CloudBurst appliance).




Community

Lies between Public and Private; a group of organisations (within the public
sector for example) with similar requirements share a single Cloud

Similar to externally hosted Private Cloud, but with more than one tenant

By service provided




IaaS


Infrastructure as a Service

H
ardware related services using the principles of cloud computing

A platform for running virtual servers, either vendor supplied images or
created by the customer.

Database services, IP addresses and persistent storage may also be offered.

Normally billed by CPU cycles, storage usage, network traffic etc.

Amazon’s EC2 and S3 are good examples.




PaaS


Platform as a Service

A development platform in the Cloud, may be thought of as a virtual
web/application server.

Billed as for IaaS.

Google’s App Engine, Microsoft’s Azure, Salesforce’s force.com.




SaaS


Software as a Service

Email, Office, CRM etc. services offered on a pay
-
per
-
user basis.

Microsoft’s BPOS and Web Office, Google’s Google Docs and Gmail, IBM’s
LotusLive, Salesforce’s CRM solution.

Infrastructure as a Service




Originally created by Amazon for their own e
-
commerce site, it is now a major revenue earner for
the company with a estimated margin of around 50%.




Powers many Facebook games as well as organisations such as Netflix, NASA, Autodesk, NASDAQ
and the New York Times.




Many different components are available, the two most important are the EC2 (Elastic Computing
Cloud) and the S3 (Simple Storage Service,
99.999999999% redundancy
).




Virtual private server instances known as Amazon Machine Images are run on EC2. Linux, Solaris
and Windows Server virtual machines are available.




S3 or Amazon’s database services (e.g.
MySQL
, Oracle) can be accessed by the AMIs or by external
applications



The AWS API has become something of a standard,

with GPL products such as Eucalyptus and NASA’s Nebula

using the API.




A high performance option is available














At its most basic, an application running on the AWS
may consist of a virtual server running on the EC2,
storage in the S3 and perhaps a message queue running
on the Simple Queue Service (SQS).

Since 1999, one of the early Cloud players.
Originally a Internet based CRM application; still
has a focus on customer relationships and service
but has diversified into both Platform and
Infrastructure offerings.




Tightly integrated with Social Networks such as
Facebook and Twitter.




Tightly integrated with the Google Data APIs


can access Google’s Calendar, Docs and Gmail



directly.






Integration with Outlook via Outlook Connector; does not yet support Windows 7 or
Office 2010 and no resolution date given.




Sales support, service support and collaboration applications available.




Sector specific versions of applications available.




Salesforce AppExchange is a marketplace for Salesforce applications and services.


Contains 965 apps and 292 services at the present time.

Agility



implementation of new services no longer needs to deal with hardware issues. Services can be brought on
-
line
quicker.


Cost



capital costs are converted to operational costs and can follow usage as capacity need not be purchased until needed.
Economies of scale can be leveraged in a shared Cloud as common services


such as backup & restore, virus searching, fail
-
over


are also shared.


Device
and
location

independence


connection to the Internet allows services to be accessed anywhere there is a
connection and using any device with a suitable client.


Multi
-
tenancy



a key characteristic of the Cloud, enables sharing of resources and costs across a large pool of users.


Reliability



is improved if multiple redundant sites are part of the Cloud infrastructure.


Scalability



also part of the Cloud infrastructure but may require attention to parallel programming when developing
applications to take full advantage of the platform.


Security



issues as well as advantages. The Cloud vendor is able to devote resources to security that many customers can
not afford, logs and audits are standard. However there are legal issues related to the physical location of data (e.g. the
Patriot Act for data hosted in the USA), as well as more normal security issues. Systems are no longer behind a corporate
firewall, extra complexity increases risk of errors, multi
-
tenancy allows for the possibility of data becoming available for
other organisations, etc. Data security is common area of concern for potential customers.


Metering and quota management



normally part of the Cloud infrastructure, vendors track space, CPU cycles etc for billing
purposes


SLA



same issues apply here as for ‘old fashioned’ outsourcing; total uptime, solution time in the case of problems, etc.

Resource sharing allows the emergence of the Cloud model



On Demand
computer resources


pay
-
per
-
use billing model



Little or no capital expenditure required



Long term commitment is not necessary



Computing as a commodity


buy computer resources on an open market


New or increased risks need to be dealt with



Security



Is my data safe from hackers or spies? (Is it any less safe than outsourced data?)



Should I keep a local copy or backup of my data?




Availability



What happens if the Internet is down?



What happens if my vendor is taken over or goes bankrupt?




Standards and interoperability



Will I become locked in to a specific vendor?



Can I use different vendors/platforms in my cloud solution?