Electronic queue into space

What distinguishes a genuine IT specialist? What lies behind information technologies? Natalya Goliaeva asked the director of the Information Technology Laboratory of the United Institute of Nuclear Research, Doctor of Technical Sciences Vladimir Korenkov - and also talked with him about grid networks, cloud structures and their scientific, engineering and social applications.

Vladimir Korenkov: System administration and web design require special training, and not always you need to have a higher education for this. But, on the other hand, to manage large IT infrastructures (supercomputers, grid systems, cloud environment), high-level system administrators are required, which are not yet prepared at universities. Our laboratory is engaged in a wide range of difficult problems, so we need high -level specialists who can adapt their knowledge to solve different problems. For example, in the shortest possible time, create a module in a new programming language, configure a data storage system or create a cloud environment. Information technologies are changing very quickly, so you need to study constantly.

However, young IT specialists need to be able to position themselves correctly. It often happens that weak specialists easily find high-paid places and become leading employees in Moscow banks and IT companies. This is rather a trouble of strong - it means that they cannot show themselves. By the way, in Dubna there are still few sites where young people could show their capabilities.

Maybe students are taught somehow not so, since those who manage well cannot find a job, and the clutches of the doubles are well settled in life?

In today's situation, there is a large proportion of the EGE guilt. Informatics in Russia made a subject for graduates for choice, so only 7 percent of schoolchildren hand it over. In schools, there are not enough computer science teachers. The low level of her teaching is supported by the backwardness of the technical base of schools. Many school principals do not even advise the children to pass the exam in computer science, orient them to choose natural science. Parents also push children along the path of least resistance. The exam in natural sciences surrenders a large percentage of graduates. With the result of the exam in science, graduates of schools cannot enter IT specialists, since for this you need to take the exam in computer science. As a result, the universities are short -range in specialties related to information technologies. There is no competition. In general, in the field of IT, the main problem of today. Universities do not have time behind the development of information technology, and experts are reluctant to teach at universities. I often have to use an administrative resource so that good experts teach at the university. Anyone can assemble a computer or create a simple information site today. And to learn how to introduce modern cloud computing or developing parallel algorithms is an too difficult task for many.

Do you mean grid technologies?

And grid technology too. Grid and the cloud are methods of organizing distributed calculations. The architecture of the Grid -network created for the Great Hadron Collider in the Cern is different from the cloud structure, but in their development there is a clear tendency to merger.

And what is the difference between the Cernovskaya grid network and the cloud?

Let me remind you that both the grid and the cloud are most often built on the basis of the allocated resources (clusters, storage systems, supercomputers, data centers). Although there are also projects for uniting individual private users in a single grid environment for joint calculations (in this case, we are talking about unbearable resources or voluntary calculations). Resources are combined to create a large computing power - that is, to use the highest possible amount of processor memory and reduce calculation time due to the implementation of calculations in parallel mode.
Grid for the Great Adron Collider and other serious tasks uses the allocated resources, and not the resources of individual user machines. Grid infrastructure provides resources only to virtual organizations (user associations that are engaged in common tasks, for example, CMS, Atlas, Alice, Biomed, Fusion, Panda), not individual users.
The grid network of a large Adron Collider has a multi -level hierarchical structure. Its main center (it is called Tier-0) is located in Cern, in Geneva. From it, several rays diverge down to the centers of the next level-Tier-1. There are only eleven such centers in the world. (Tier-1 Twelfth Center is now being created jointly by JINR and the Kurchatov Institute. This will be the first center of this level in Russia). From each of the Tier-1 centers there are rays to centers of an even lower level-Tier-2. And so on. At zero - the highest level - primary information is collected and stored. At the first level, a full copy of the information collected at the zero level is stored. And at the second and lower levels, information is not stored, but only processed - calculations and analysis are made.

Since the participants of the grid structure are endowed with various priorities, in practice it turns out that a small number of users can collect and use all network resources. While customers with a higher priority use the network, the rest should wait for their turn. Therefore, the tree-like grid structure is compared with a sand clock. The top of the watch is zero and the first level of the grid network. The lower flask is the second and lower levels. Information from the highest levels when transmitting down should pass through a narrow neck. And this is the problem of the grid structure, which requires solution by improving technologies. First of all, it is necessary to increase the rate of exchange of information from the current 10-20 Gb/s to 100 Gbit/s (in the JSI we plan to achieve this by 2015).

Are the clouds organized differently?

Cloud infrastructure is most often created on the basis of one or more data centers using virtualization and providing various services to both organizations and individual users. Models for providing cloud services can be different: IAAS -infrastructure as a service, PAAS - platform as a service, SaAS - software as a service.

What do these abbreviations mean?

The SaAS model (from the English Software As a Service) allows their customers to use the applied software of the provider. Access to it is provided using a browser or program interface. Each of the users of the cloud has the right only to a small part of the total pool of the resources of the cloud. Therefore, one user cannot cause a general lack of resources. So in cloud computing, an illusion of an endless computing space is created.

In the PAAS model (from the English Platform as a Service), the main service is the use of cloud infrastructure as a platform for placing the basic software of the client and applied applications to it.

The cloud model of IAAS (from the English. Infrastructure as a Serge) gives the client all its infrastructure, where you can independently manage the network, processing and storage of data.

Grid technology and clouds is better than supercomputers?

Cheaper with the same computing efficiency. The use of the grid network is sometimes compared using an electrical outlet, where it is also easy to connect any electrical appliance on how to include a personal computer to the grid network. However, experts in the field of information technology predict the rapid inclusion of supercomputers in cloud and grid structures.

What information will the Tier-1 center be stored, which is created by JINR and the Kurchatov Institute?

The grid network is organized for processing a huge array of data that continuously comes from the detectors of a large adrone collider. The CMS experiment will be stored in JINS, and the ATLAS detector detector data at the Kurchatov Institute. I must say more precisely-the United Institute will create a center for storing the data of its participating countries. And our physicists, thanks to this possibility, will be able to more actively participate in the analysis of the results of experiments of the Great Adron Collider. Work experience in creating the center of the first grid network is very useful to us in the development of information infrastructure for scientific megaprojects of Russia - Nica collider in Dubna and the peak reactor in Gatchina. This is a new technological level - to accept, store and process information with high speed and reliability. This knowledge is very lacking in Russia in many high -tech areas, for example, in the space industry or in medicine.

It seems to me that in earthly conditions, working with information is somehow not very well thought out. Take at least the sensational project of electronic queues in clinics - in the form as it was presented to us, it is no different from the usual self -recording procedure for the doctor. Only self -reference is practically free, and equipment for an electronic queue costs frantic money ...

Yes, the electronic queue system is not impressive ... In my opinion, all this happens because outdated information technologies are used. And for development, we need to adopt the experience of creating the information infrastructure of the Big Adron Collider. It must be applied, inter alia, in healthcare to create a medical data database. But Gazprom, Rosatom, and the MiG, and Sberbank air corporation are concerned: information processing requires a serious data protection system, including personal ones. Although the Boeing and Intel corporations did so well with data protection that they are not afraid to involve Russian engineers and programmers in their development. It seems to me that here it is not so much the fear of our top managers to lose important information as unwillingness to work at a modern level.