Any corporate IT landscape consists of many applications, most of which have their own databases. These databases store information objects that represent business objects, events, and phases of business processes. Many business process objects are “reflected” in several databases at once: for example, a piece of equipment of an industrial enterprise is described from different points of view in accounting systems, repair and maintenance management, production management, etc.
In order for business applications that automate different business processes to somehow work together, they need to be integrated: implement products of the class MDM (Master Data Management, master data management system) and ESB (Enterprise Service Bus, corporate service bus), which allow somehow manage the exchange of information between a variety of multi-platform solutions. Those who have been involved in such integration are well aware that this is a long, difficult and thankless job.
But what if there is a way to get rid of all the integration problems at once? Such a "magic bullet" exists, and it is called - data-centric architecture. Its main idea is to make data, not business applications, at the center of corporate IT architecture. This principle is outlined in the Data-Centric Architecture Manifesto and in The Data-Centric Revolution: Restoring Sanity to Enterprise Information Systems .
Imagine that a company has a single virtual data warehouse in which each business object or event exists in a single instance. For clarity, you can imagine that the idea of the MDM system has been brought to a logical full embodiment, and it is MDM that is the repository of all corporate data; business applications do not have their own DBMS and only work with data objects from MDM. The advantages of this architecture are obvious:
The need for integration procedures is eliminated once and for all.
Data storage costs are reduced by eliminating multiple copies of each business object across different systems.
Analytics and decision support are simplified, since now, to build any analytical slice, you no longer need to extract and glue data from different systems for months - they are always at hand.
, -.
- , . -- -.
« - », . - , -?
, , , -, , , -. , — , , - -. , , .
, , , «» ( ) , — - . , - .
«» (corporate data cloud) (data lake)? — , . data lake , - - , - « — ». , ...
— , . , , , ; -, . , ; , . (. OWL W3C).
, , . ( ), , , . , , .
, (corporate knowledge graph), .
, - ? :
. — , «» .
() . , . . .
API , REST, GraphQL, SPARQL.
.
.
(data provenance), (data quality), .
. - , — , , (. SHACL SHACL Advanced Functions). - low code: , - , -, , « » , .
Platforms that meet these requirements already exist and are used both in the Russian and foreign markets.