Prologue
Forbes, a globally renowned journal, and Gartner, the world’s leading research institute, are paying attention to graph databases as a core technology that will empower the business and work environment. In other words, graph technology is proven to be effective in analyzing and utilizing complex relational data in contrast to data management methods that have been used in the past. Corporations and research agencies around the world are rushing to implement graph technology to improve their business. Manufacturing industries, in general, considers data provenance as one of the cores of data management. Data provenance refers to recording details of origin, changes, and details supporting the authenticity of data.
This use case will introduce a shipyard company based in South Korea (hereinafter referred to as Company D), which was looking for a way to improve data history management and search services using graph technology while also maintaining the existing relational database. Company D leads the future of the offshore industry with hardware (ships, submarines, offshore plants, etc) technology and AI-based software technology.
A ship is manufactured from many raw materials, sub-components, and parts, etc. The list of these materials is also known as ‘bills of material” (BOM). In a shipyard that develops dozens of ships and other technologies, there is bound to be an immeasurable amount of BOM data on these materials.
Employees would search for either the model name, material number or blueprint number to retrieve the desired BOM information. Company D used Oracle, one of the representative relational databases, prior to adding Bitnine’s AgensGraph into their system. The use case will now introduce why and how Company D efficiently acquired data provenance in their data management system by utilizing the advantages of both the relational DB and graph DB.
Challenges of ship manufacturing company
Manufacturing a ship takes a long time. During this period, related parts or materials and information regarding the subcontractors must be frequently updated.
Company D has been using Oracle as data storage and management. One of the biggest strengths of RDB is its capability to manage millions of data of several ships in a table format. However, RDB has its weakness when it comes to processing speed. When performing a query in an RDB, the more data and users there are, the more time it takes to perform relational operations in the process of joining tables. Since more than 100 million data are included in tables, the modeling structure is difficult to see at a glance. This is why Company D struggled to analyze the relationship and impact between data with the existing relational system.
Another reason for changing the design of the existing data analytics system was to identify and analyze the relationship between data to see how a change of entity can affect another entity. For example, if a certain component had to be changed for design purposes, there might be another component that needs to be changed due to the previous component being incompatible with the newly changed part.
In order to develop a new data provenance management system, Company D found AgensGraph, which was faster in querying and analyzing than the existing system and highly compatible with relational data.
Achieving data provenance with graph modeling
AgensGraph is a multi-model graph database built on fork of PostgreSQL, an open-source relational database. With the help of graph experts from Bitnine, Company D developed a new unique structural system. Manufacturing data is stored in the existing relational database and viewing and analyzing data modeling is performed with AgensGraph-implemented search service.
In GDB, the relationship between nodes is connected by an edge, which greatly reduces the query time. Refer to the following mock simulation project of Company D, which shows how each data can be searched, tracked, or analyzed.
Searching project model
For starters, suppose there is a project called 2358, which can be a project number of a ship model. When searching the model of the materials that make up the ship (eg.104CM-P, 104UM-P…), the product name of the model can be viewed as well. Alternatively, when searching the product name as shown below, one can get access to a model, related project, or stage of the project.
Searching product name of the model
There is a step in how each and every product model is prepared. By analyzing a certain model, one can see which model belongs to which project, how much the ship blueprint has undergone changes, and which materials are replaced.
Searching steps of product model
A use case with advantages of RDB and GDB combined
As shown in the images above, AgensGraph can express relational table data as nodes in a graph data model. By implementing AgensGraph into their new data management system, Company D is now able to take a step closer to the future innovation of the maritime industry.
The newly developed data provenance management system has kept the advantages of each database. The advantage of a graph database is known for its query speed and relationship between nodes, while the relational database is efficient in storing ever-increasing log data.
The knowledge platform makes it easier to analyze the relationship and impact between data, and it is also capable of keeping track of the history of various service stages. The data provenance management has evolved further with a graph database, but that does not mean the existing relational database is forgotten. Just like how NoSQL stands for ‘not only SQL’, this use case shows a primary example of AgensGraph applied together with a relational database, not replacing it. Along with Oracle, AgensGraph can be implemented with other relational databases such as PostgreSQL and MySQL, etc.
Although this use case introduces a system applied in ship engineering, with the right adjustments, AgensGraph can be applied to any manufacturing field such as semiconductor equipment, automobiles, and chemical products. In the future, this type of hybrid technology will be expected to be the new normal, whether it is for different systems, solutions, or databases.
If you have any questions, feel free to contact us.
Thank you! 😀
Download AgensGraph ! Contact us