Introduction to Cloud Computing
Like all IT professionals, I am confronted with this fundamental issue just about every day. If you follow the different newspapers, you will have noticed a few comments a bit annoyed from time to time, but let’s try to be constructive and to put some landmarks…
A story of words
First of all, do you know that the funny terms that pop up every 2 years like Cloud Computing, Service Oriented Architecture, Big Data and Internet of Things are not necessarily from the same boxes or the same pharmacies and in the case of “Cloud Computing”, paternity is not obvious? Nevertheless, after a few months or years, the keyword begins to spread and some bearded beggars undertake to describe the ontology necessary for this new paradigm.
In the case of cloud computing, academics on the planet, pioneers at Amazon, Salesforce, Google, and Microsoft have begun to write about their definitions, advice.
After a few years, it is starting to fall and the various standardization bodies are giving birth to public definitions. In our case, the National Institute of Standards and Technology (NIST) published this small booklet in 2011.
In 2014, ISO is splitting a small breviary “IT-Information Technology Cloud-Overview and vocabulary “also known by the small name of ISO / IEC 17788: 2014.
The main principles
For the future, I will use NIST definitions that have the merit of being public, concise, proven and flexible (the NIST states that the paradigm is constantly evolving).
“Cloud computing is a model to enable access to demand (important criterion) by means of a network (such as the Internet) to a group of resources ( E.g., networks, servers, storage, applications or services) that can be quickly provisioned and released with minimal management effort or minimal interaction with the vendor.
This model presents 5 essential characteristics, 3 models of services and 4 models of deployments “.
I emphasize the 5 essential characteristics:
- “On-demand self-service”: A “consumer” of the service must be able to provision the service according to its needs without mandatory interaction with a human.
- “Broad network access”: Resources must be accessible across the network using “standard” mechanisms (regardless of a heavy or light client).
- “Resource pooling”: The resources must be aggregated to serve different clients according to a “multi-tenant” model.
- “Rapid elasticity”: Resources must be allocated and deallocated quickly on an elastic model.
- “Measured service”: The consumption of resources must be measured and monitored using a mechanism that gives transparency to consumers and service providers.
The 3 models of services described by NIST are IaaS, PaaS, and SaaS. We’ll come back to that.
The 4 models of deployments are often misinterpreted. We distinguish:
- “Private Cloud” – The private cloud: In this model, the infrastructure is provisioned for the exclusive use of a single organization that can share this infrastructure between its various subsidiaries.
- The infrastructure may be owned by the organization itself, a third party or a combination of both, and operated locally or hosted by a third party.
- “Community Cloud” – The Community Cloud: In this model, the infrastructure is provisioned for the exclusive use of a customer community that has been organized, for example at a level such as European aeronautics.
- “Public Cloud” – The public cloud: In this model, the supply of infrastructure is open to “public”.
- “Hybrid Cloud” – The hybrid cloud: In this model, we have a combination of previous offers that remain independent but those share standards allowing interoperability, the case of use often presented is the overflow of a private cloud to the public cloud in case of peak load.
3 Models of Services (IaaS, PaaS, SaaS)
The NIST document, therefore, describes 3 models of services (IaaS, PaaS, SaaS) that are the object of many misconceptions, confusions or other attempts of smoky marketing positioning. The key concept in service models is the very logic of “as a Service”, which takes up the features detailed above. We buy access to a resource and not the resource itself.
IaaS: Infrastructure as a Service
In the case of the “infrastructure”, the cloud provider will make the infrastructure components available (servers, storage, network, etc.) that are mostly virtualized, enabling it to ensure the 5 Essential characteristics for it to be of the true “cloud”, but attention is not always the case! That’s why you should read the small lines of contracts, or better ask your buyers and your favorite lawyers to peel out the small lines of contracts for you. This makes it easier to understand the logic of “as a service”, which leads to spending more time on contracts and less time on infrastructure. Suppliers now sell access to resources, Operated according to standards, and not the resources themselves that you must operate or have operated at home. It’s the magic of Cloud Computing. DRHs, DAFs and DSIs are delighted, because your box needs less staff (bearded and hard) to configure, install and monitor applications, less CAPEX (expensive hardware, complicated to dampen and life cycle enough Short), and the DSI reacts faster than its shadow to provide a complicated infrastructure for a VIP project (any resemblance).
In this service model, the provision of virtualized servers usually comes with the operating system and hardware required to administer (supervise, backup, etc.) the server, but not more. It can be noted that our preferred OS is’
SaaS: Software as a Service
The “Software as a Service” is a service model where the application can be accessed using a standard client (thin client, web browser) and provisioned on demand. This model is widely available to the general public, sometimes even “free”, the emblematic example being the online suite proposed by google (gmail, google docs, google drive, etc.). A classic misunderstanding involves confusing the proposed suite (features) with the Google SaaS service model. What makes Google offer a “Cloud Computing” offer is how you can enroll and then access the solutions. As a result, setting up a small server at home to store your data and that of your other with a solution called “Personal Cloud”.
Beware, many historic hosted applications, on demand, have been dressed in “SaaS” sauce with the advent of Cloud Computing. Again, the careful reading of the small lines of the contract, the audit of the infrastructure, its performance or housing conditions may prove necessary before taking the step. Nor should we neglect the cost of doing business,
PaaS: Platform as a Service
The “Platform as a Service” is a model where the provider provides a “platform”, which consists of one or more infrastructure components and/or pre-configured software (in addition to the operating system). This platform will allow the customer to develop his own code. It is the most complex model to be apprehended, and the solutions proposed by the suppliers are extremely varied. Between the definition of NIST (2011) and now, many solutions have emerged among large suppliers (e.g., Google App Engine, BlueMix at IBM, OpenShift at RedHat)
Many intermediate offers (MBaaS, DBaaS, iPaaS, aPaaS, FaaS) appeared and surfed on the “as a Service” wave. Most can cling to one of the 3 large models, avoid adding more (“PAAS ++”, this is not a good idea).
I hope you like it!