The growth of computing power and the development of x86 platform virtualization technologies on the one hand, and the spread of IT outsourcing on the other hand, have led to the concept of utility computing (IT as a public service). Why not pay for IT in the same way as for water or electricity - exactly as much and exactly when you need it, and no more.
At this moment, the concept of cloud computing appeared - the consumption of IT services from the "cloud", i.e. from some external pool of resources, without caring about how and where these resources come from. Just like we don't care about the infrastructure of water utility pumping stations. By this time, the other side of the concept had also been worked out - namely, the concept of IT services and how to manage them within the framework of ITIL / ITSM.
A number of definitions of clouds (cloud computing) have been developed, but this should not be taken as the ultimate truth - it is just a way to formalize the ways in which utility computing is provided.
- "Cloud computing is a distributed data processing technology in which computer resources and capacities are provided to the user as an Internet service" Wikipedia
- βCloud computing is a model for providing convenient network access to a shared pool of customizable computing resources (e.g., networks, servers, storage, applications, and services) on demand that can be quickly provisioned and provisioned with minimal management effort or minimal intervention. Service Provider" NIST
- βCloud computing is a paradigm of providing network access to a scalable and flexible pool of distributed physical or virtual resources, provided in a self-service mode and administered on demandβ ISO/IEC 17788:2014. Information technology - Cloud computing - Overview and vocabulary.
According to NIST, there are three main types of clouds:
- IaaS - Infrastructure as a Service - Infrastructure as a Service
- PaaS - Platform as a Service - Platform as a Service
- SaaS - Software as a Service
For a very simplified understanding of the difference, let's consider the Pizza-as-a-Service model:
NIST defines the following necessary features for an IT service to be considered a cloud service.
- Broad network access - the service should have a universal network interface that allows connection and use of the service by almost anyone with minimal requirements. An example - to use the 220V electrical network, it is enough to connect to any outlet with a standard universal interface (plug), which does not change whether it is a kettle, a vacuum cleaner or a laptop.
- Measured service - A key characteristic of a cloud service is service measurability. Returning to the analogy with electricity - you will pay exactly as much as you consumed with a minimum granularity, up to the cost of boiling a kettle once, if you were in the house once and drank a cup of tea during the whole month.
- Self-configuring services on demand (on demand self service) - the cloud provider provides the customer with the ability to intelligently configure the service, without the need to interact with the provider's employees. In order to boil the kettle, it is absolutely not necessary to contact Energosbyt in advance and warn them in advance and obtain permission. From the moment the house is connected (a contract is concluded), all consumers can independently dispose of the provided power.
- Instant elasticity (rapid elasticity) - the cloud provider provides resources with the ability to instantly increase / decrease capacity (within certain reasonable limits). As soon as the kettle is turned on, the provider immediately releases 3 kW of power to the network, and as soon as it is turned off, it reduces the output to zero.
- Pooling of resources (resource pooling) - the internal mechanisms of the service provider allow you to combine individual generating capacities into a common pool (pool) of resources with further provision of resources as a service to various consumers. Turning on the kettle, we are least of all worried about which particular power plant the power comes from. And all other consumers consume this power together with us.
It is important to understand that the characteristics of the cloud described above are not taken from the ceiling, but are a logical conclusion from the concept of utility computing. And the public service must have these characteristics within the framework of the concept. If one or another characteristic does not match, the service does not become worse and does not become βpoisonousβ, it just ceases to be cloudy. Well, who said that all services should?
Why am I talking about this separately? In the past 10 years since the NIST definition, there has been a lot of controversy about "true cloudiness" according to the definition. In the United States, the wording βcorresponds to the letter of the law, but not the spiritβ is still sometimes used in the judicial sphere - and in the case of cloud computing, the main thing is the spirit, resources for rent in two mouse clicks.
It should be noted that the above 5 characteristics are applicable to the public cloud, but when moving to a private cloud, most of them become optional.
- Universal network access (broad network access) - within a private cloud, the organization has complete control over both generating capacity and consumer customers. Thus, this characteristic can be considered as automatically performed.
- Measured service is a key feature of the utility computing concept, pay-as-you-go. But how do you pay organizations to themselves? In this case, there is a division of generation and consumption within the company, IT becomes a provider, and business units become consumers of services. And the settlement takes place between departments. Two modes of operation are possible: chargeback (with real mutual settlements and the movement of finances) and showback (in the form of reporting on the consumption of resources in rubles, but without the movement of finances).
- Self-configuring services on demand (on demand self service) - within the organization there may be a common IT service, in which case the characteristic becomes meaningless. However, if you have your own IT staff or application administrators in your business units, you need to set up a self-service portal. Conclusion - the characteristic is optional and depends on the structure of the business.
- Instant elasticity (rapid elasticity) - within the organization loses its meaning due to the fixed set of equipment for organizing a private cloud. It can be used to a limited extent in the framework of internal mutual settlements. Conclusion - not applicable for a private cloud.
- Pooling of resources (resource pooling) - today there are practically no organizations that do not use server virtualization. Accordingly, this characteristic can be considered automatically performed.
Q: So what is your private cloud anyway? What does a company need to buy and implement in order to build it?
Answer: A private cloud is a transition to a new administrative model of IT-Business interaction, which consists of 80% of administrative measures and only 20 of technologies.
Paying only for the resources consumed and easy entry, without having to bury several hundred million oil in capital expenditures, has created a new technological landscape and the emergence of billionaire companies. For example, modern giants Dropbox and Instagram appeared as startups on AWS with zero infrastructure of their own.
It should be emphasized separately that cloud service management tools are becoming much more indirect, and the key responsibility of the IT director is the selection of suppliers and quality control. Let's take a look at these two new responsibilities.
Appearing as an alternative to classic heavy infrastructure with its own data centers and hardware, clouds are deceptively light. It is easy to enter the cloud, but the issue of exit is usually bypassed. As in any other industry, cloud providers strive to protect business and make it harder to compete. The only serious competitive moment arises only with the initial choice of a cloud service provider, and then the provider will make every effort so that the customer does not leave him. Moreover, not all efforts will be directed to the quality of services or their range. First of all, it is the delivery of unique services and the use of non-standard system software, which makes it difficult to switch to another provider. Accordingly, when choosing a service provider, it is necessary to simultaneously form a transition plan from this provider (in fact, a full-fledged DRP - disaster recovery plan) and think over the architecture of data storage and backups.
The second important aspect of the new responsibilities of the CIO is the quality control of services from the supplier. Almost all cloud providers comply with SLA according to their own internal metrics, which can have an extremely indirect effect on the customer's business processes. And accordingly, the implementation of your own monitoring and control system becomes one of the key projects when transferring significant IT systems to a cloud provider. Continuing the topic of SLA, it should be emphasized that the vast majority of cloud providers limit liability for non-fulfillment of SLA to a monthly subscription fee or to a share of the payment. For example, AWS and Azure, when the availability threshold of 95% (36 hours per month) is exceeded, will make a 100% discount to the subscription fee, and Yandex.Cloud - 30%.
And of course, we must not forget that clouds are not only performed by Amazon-class mastodons and Yandex-class elephants. Clouds are also smaller - the size of a cat, or even a mouse. As the CloudMouse example showed, sometimes the cloud just takes and ends. You won't get compensation or discounts - you won't get anything but total data loss.
In view of the above problems with the implementation of IT systems of high class business criticality in cloud infrastructures, the phenomenon of "cloud repatriation" has been observed in recent years.
By 2020, cloud computing has passed the peak of inflated expectations and the concept is on its way to a ditch of frustration (according to the Gartner Hype Cycle). According to research ΠΈ up to 80% of corporate customers return and plan to return loads from the clouds to their own data centers for the following reasons:
- Improve availability/performance;
- Reduce costs;
- To comply with IS requirements.
What to do and how everything is "really"?
There is no doubt that the clouds have come in earnest and for a long time. And every year their role will increase. However, we do not live in the distant future, but in 2020 in a very definite situation. What to do with the clouds if you are not a startup, but a classic corporate customer?
- Clouds are primarily a place for services with unpredictable or pronounced seasonal load.
- In most cases, services with a predictable stable load are cheaper to maintain in your own data center.
- It is necessary to start working with clouds with test environments and low-priority services.
- Considering the placement of information systems in the cloud begins with the development of a methodology for moving from the cloud to another cloud (or back to your own data center).
- Placing an information system in the cloud begins with the development of a backup scheme for the infrastructure you control.
Source: habr.com