RSS
Логотип
Баннер в шапке 1
Баннер в шапке 2
2018/02/12 18:55:00

Computing without servers - what stands behind this mysterious concept?

With development of cloud services computing becomes more and more similar to utility, and the physical server framework which executes calculations ceases to interest consumers.

Content

The subject "computing without servers" (it is so possible to translate serverless computing) is in the left part of a gartnerovsky curve of technology trends of 2017. Let's notice what from reasonable precaution of Gartner uses more modest term Serverless PaaS. An important circumstance - Serverless PaaS position mentioned by a circle of green color assumes movement of serverless computing of the stable right side in the next several years.

According to analysts of Gartner though the technology is in a germ, its development and adaptation to practical use will demand very limited time frame.

The name Serverless was offered in 2012 by Ken Fromm in article "Why The Future Of Software And Apps Is Serverless"[1].

The dissociation of opinions on serverless computing is reflected by the picture given below.

Speaker: "Computing without servers is cool!"

Skeptical listener: "Why by him not to call this case by just remote procedure call?"

Such judgments reflect the keen interest in serverless computing which is followed as usual in such situations, both excessively optimistic, and excessively pessimistic estimates. From ourselves we will add, as stated in the famous parable about the judge, his wife and two women dividing the child: "And you the woman are right too", namely, there is the third opinion and it correct, serverless computing too - abruptly, but not really, and of course it is not a remote call of procedures.

New approach to providing PaaS services in the form of functions

Speaking about serverless computing, it is necessary to understand that actually not to do without physical virtual servers under no circumstances, means it is not about failure from "iron", and about new approach to providing PaaS services in the form of functions. This service was called by FaaS (Function as a Service).

Understand such category of cloud services which provides the platform for development and accomplishment of applications without any encumbrance connected with the supporting infrastructure as FaaS. The user does not see servers, in this sense infrastructure can be considered besserverny (serverless). This position reminds the kid who, closing eyes palms, thinks that he is invisible.

For the first time FaaS services were implemented in 2014 in[2]. Soon there were Amazon AWS Lambda services, in 2016 - Google Cloud Functions, Microsoft Azure Functions, IBM/ Apache's OpenWhisk and in 2017 - Oracle Cloud Fn.

Hook.io represents the platform with the open code supporting a hosting of microservices (microservice) and web interceptions (webhook hosting).

Microservices are a present stage in development of the well-known service-oriented architecture (SOA). Microservices represent a set of loosely coupled components which interact on network with each other for accomplishment of the purpose. Webhook is a method of development of web applications with the notification mechanism about the events which are taking place on web pages.

Serverless computing is a kind of cloud services

In English-speaking Wikipedia the term serverless computing is absolutely fairly referred to category misnomer. This word, surprising on accuracy, unfortunately, has no adequate translation into Russian. Approximately it can be interpreted as "the wrong name", but rather it means "popular belief with which it is necessary to be reconciled". It is possible to give the web services which do not have relations to Web, or "switching factories", in general as an example of similar misnomer as not connected with factories.

However, it is necessary to be surprised to the fact that very good and perfectly corresponding term is. It is utility computing, but for the unclear reason it constantly remains unclaimed though it is known very long time ago. In 1961 John McCarthy, the author of the term "artificial intelligence" and the inventor of the LISP programming language, told:

File:Aquote1.png
If my idea of computers of the future is correct, then sometime computing will be organized as any other utility, as telephony, for example. And this utility computing will become a basis for new and extremely important industry
File:Aquote2.png

John McCarthy

If McCarthy told only it, he would be already worthy memories of descendants, but he is better known that he was fond of artificial intelligence and piled as we understand now, it is a lot of nonsense. So in 1966 as a task to students for the summer he suggested to create such robot who could assemble the color television receiver from the corresponding set of parts. The task for summer — is not less and no more... Such is there was his understanding of complexity of a problem of AI, immense optimism from here. Eventually the TV was set up there where it was conceived, but, alas, students assembled it manually.

With other utilities it is simpler and simpler. In them it is obvious as as it is transferred to the consumer. And what reasonable form of provision of services of computing? The most banal form – IaaS. The standard server framework or DWH is provided to you in use.

However, historically "a two-humped abbreviation" became the first SaaS, grown from the services offered by ASP providers (Application Service Provider). Then there were DaaS, DBaaS, PaaS … - in a word XaaS, i.e. "anything as services". All this is no other than search of forms for representation of those services about which John McCarthy wrote half a century back.

The form of providing software as services – Software as a Service (SaaS) for years of the existence seriously changed. It has three persons – containers (CaaS), applications (PaaS), to them functions (FaaS) increase. It would be possible to stop on it, but to marketing specialists to liking serverless computing. So FaaS is that other as the next step towards Utility computing.

In the current implementations a number of restrictions is detected in platforms of FaaS: complexity of the organization of parallel computings, considerable delay. Nevertheless FaaS can be considered as the next step towards the full Utility computing system and to express confidence that through some time the next step in the same direction will be taken.

Evolutionary steps in development of IaaS and SaaS in the Utility computing direction

The qualitative difference of FaaS from PaaS is that a subject of service is not some monolithic complete web application, and the loosely coupled system of components distributed on a cloud allowing to create architecture which work is subordinated to events (event-based architecture). Such system is not tied to any physical servers and not limited on scaling (auto-scalable).

Informative introduction to FaaS can be found in the material "Serverless Architectures" of Martin Fowler, the famous popular writer of computer knowledge[3].

The GitHub-page[4] will be useful to developers].

Notes