Cloud DBMS (databases) Cloud Database Management Systems, CDBMS

2020: Market leaders of cloud DBMS are called

In November, 2020 the Gartner analytical company published results of a research of the market of cloud DBMS. Experts made a so-called magic quadrant in which market leaders are listed.

The following companies are carried to them:

• Amazon (more precisely, cloud division — AWS);
• Microsoft;
• Google;
• Oracle;
• SAP;
• IBM;
• Alibaba (more precisely, Alibaba Cloud);
• Teradata.

Each producer has merits and demerits. For example, Amazon it is called the supplier of cloud DBMS with the best indicators of availability and reliability. To pluses are carried the high performance of cloud infrastructure, and to minuses — integration problems with services and creation of the different services used for the solution of numerous tasks.

Magic quadrant, market leaders of cloud DBMS

At Microsoft experts note the correct vision of a cloud ecosystem of data, width and depth of a product portfolio and also possibility of migration in Azure, connected that many companies in the world use already software of Microsoft.

At the same time at corporation the cloud ecosystem, and transition to Azure did not ripen yet, according to clients, is more expensive in comparison with local applications launch.

Analysts consider that by 2022 75% of databases will be developed in the world on cloud platforms or it is transferred to them. Only for 5% of bases the possibility of the return transfer on own infrastructure will be considered.

Their use for analytics and distribution of the SaaS model is the main reasons of transfer of databases on cloud platforms. First of all transfer the information storages, lakes of data and the other systems focused on problems of analytics, artificial intelligence and machine learning to a cloud. Growth in sales of systems for installation on own infrastructure is noted too, but it is more connected not with new installations, and with increase in prices and forced updates.^[1]

2018

According to analysts of Gartner, in 2018 the world market of DBMS in general increased by 18.4%, to $46 billion, and 68% of growth were the share of cloud DBMS. And by 2022 75% of databases or will already become cloud, or will stay in the course of migration in a cloud. It is expected that much interest in CDBMS will be shown by the companies which are engaged in modern applications, such as machine learning and analytics^[2].

The novelty of cloud DBMS (Cloud DBMS, CDBMS), as well as other cloud software, consists in failure from a traditional "product" form of distribution. The user is saved from need to invest means in acquisition of a required set of hardware and program resources, instead he gets access to the similar resources placed in the DPCs belonging to cloud provider. At the moment two basic approaches to implementation of CDBMS are implemented.

First, it is possible to use IaaS service, to rent necessary infrastructure resources. The cloud platform allows to reserve the necessary number of copies of the virtual machine, DBMS distributed on them supporting work for time desirable to the user. At the same time the client can set DBMS elected by him on the rented virtual machines. In the second option he uses in advance prepared machine images provided by DBMS producers together with cloud provider, for example, to Oracle Database 11g EE on Amazon EC2.

The second approach to CDBMS has in a basis the scheme SaaS, "Software as a service" where as software DBMS therefore it is given own name DBaaS (Database as a Service) is used. Providing DBaaS services, providers save clients from need for any independent actions with copies of virtual machines. The provider undertakes these functions, as well as all responsibility for installation and maintenance of base. In that case the user needs only to use the ready-made solution and to pay its work. For example, Amazon Web Services offers the interface with MySQL DBMS.

The first option has well-known merits and demerits of cloud solutions. Treat advantages cost efficiency, unlimited (at least — in the theory) scaling options, lack of any problems with updating of the equipment at emergence of new requirements, big flexibility and reliability. Shortcomings — the client cannot control the data security limited to the agreement on the service level; malfunctions in data networks can lead to a work stop, placement of data and the software in DPC of provider causes affection for him.

During the work on the DBaaS model there is no need to purchase software licenses, to have staff of the specialists supporting the database the provider — in that degree in what it is written in the agreement on the service level is responsible for reliability of work of DBMS. The main lack of DBaaS — a smaller extent of control of data.

When choosing any of two options follows, first of all to proceed from scales of the problems facing the enterprise. DBaaS is more preferable to small and medium business in modern conditions, for larger there can be sufficient a lease of the equipment of IaaS, and to not numerous very large enterprises working with huge amounts of data, nevertheless it is more reasonable to place databases on own squares and on own equipment.

Well and, of course, a lot of things depend on tasks which should be solved. If to take tasks "normal", then for the choice of provider it is possible to use assessment of analysts of Gartner about the state of affairs in the market of operational DBMS (ODBMS). These are DBMS intended for the solution of a broad spectrum of corporate transactional applications (ERP CRM etc.).

2000-2010

Historically databases were under construction proceeding from idea unique at that time of vertical scaling (scale-up) assuming use of more and more powerful servers. At the beginning of the 2000th years apogee of this line of development in it the directions became several powerful DBMS (Oracle, DB2, etc.) working at multiprocessor as Unix servers. But with emergence in the same years of thin servers (height 1-2 U) and especially servers edges, vertical scaling began to give way gradually to cheaper horizontal scaling (scale-out) where server power is increased due to consolidation in clusters of rather low-power servers of standard architecture of x86. As a result of the database "changed" on the cluster configurations collected from servers of standard architecture or on the hyper convergent systems.

In about five years clouds appeared, a noticeable part of a computer landscape they became by 2010. And here it became clear that the versions of DBMS which are initially expected clusters were in advance prepared for porting on the infrastructures provided by providers in the form of cloud services.

And further multiplied by unlimited potential of scaling of server resources and volumes of the storage systems provided by cloud providers these two factors created the synergy effect which became a source of success of cloud DBMS. The first took relational databases a way of migration to clouds, soon, after overcoming some technical difficulties, not relational also joined them.

Notes