For decades, relational database management systems (RDBMSs) have been dominant. But, while still widely used, general-purpose RDBMSs are now facing competition from NoSQL, or not-only SQL, databases. These newer systems are able to address big data and IoT use cases that were simply unimaginable years ago.
The Time Series Advantage
Every day, time series data is being created by everything from smartphone apps to smart factories. It is increasingly important for companies to be able to gain insights from this data. Time series databases make it possible.
"With the rise of Industry 4.0, in which time series data has become prevalent, there is not a better database to store this data than a time series database," explains Angga Graharya Suherman, GridDB OSS Lead at Toshiba Digital Solutions Corporation. "Conventional databases such as RDBMSs were never designed to handle this kind of data and, as a result, performance bottlenecks often occur."
The GridDB Value Proposition
Toshiba has extensive experience in building mission-critical systems. Nonetheless, before GridDB, the company found that there was not a single database that could power these systems well, notes Suherman.
"We evaluated Hadoop systems, as well as relational and NoSQL databases, but none offered the best fit for our requirements," says Suherman. "We decided to build a database in-house to serve those needs. GridDB is purpose-built from scratch to fulfill the requirements of large-scale IoT and big data systems. With our expertise in building mission-critical systems, we have a holistic understanding of what is necessary for an IoT database."
Toshiba's GridDB delivers the best of SQL and NoSQL in a purpose-built, in-memory, distributed IoT database. It is the first of its kind to use a key-container model and time series functionality to provide the capabilities necessary to create cutting-edge IoT solutions.
GridDB's "memory first, storage second" design delivers as much as 12x higher ingestion rate and 25x faster read at scale than a competing NoSQL database. It scales out linearly and horizontally using commodity hardware while maintaining high performance and eliminates single points of failure through hybrid cluster management and a fault tolerant system.
Time-stamped data from IoT sensors can be handled with GridDB's time series containers (a container is similar to a table in a relational database) and special functions such as management of data retention and time series data aggregation. Unlike many other NoSQL databases, GridDB also ensures ACID guarantees for transactions at a container level while still providing high performance and system scale-out.
New Problems Require New Solutions
"GridDB is optimized for large-scale implementations with terabytes and even petabytes of data," says Suherman, who adds that its main strength is in performance, with fast ingest and querying, and scalability. Toshiba has run benchmarks against the top time series database, as well as a popular NoSQL database platform, and a widely used relational database system, and GridDB outperformed each of them, says Suherman.
In the real world, GridDB has excelled at providing the capabilities that developers need to create solutions in diverse fields, such as power generation, building management, manufacturing, transportation, retail, and more.
For example, an electric power company is using GridDB to support a wheeling system consisting of millions of smart meters. "GridDB replaced their conventional RDBMS because they stumbled upon a performance bottleneck. They required tens of times greater performance than what the existing database could handle," says Suherman.
And a leading automotive supplier for car manufacturers is using GridDB to power the company's IoT platform. "Within 3 months of the platform's introduction, they improved their factory's productivity by 6%," observes Suherman.
The Open Source Community
Just as there are new approaches to data management today with NoSQL databases adding new capabilities, there are new ways to foster improvement in database technology as well.
In 2016, Toshiba released a community edition of its scalable time series database to the open source community to allow more developers to use it in combination with other software and to encourage collaborative innovation. "Simply put, the more heads the better," says Suherman. "In addition, by opening our arms to the open source community, we can understand more of what our users actually want, and address their demands."
Enterprise or Open Source
GridDB Standard Edition (SE) is the enterprise-grade commercial version of GridDB, an in-memory NoSQL database designed for mission-critical IoT applications. GridDB SE is packaged with professional technical support, frequent maintenance, bug fixes, patches, and updates. GridDB Community Edition (CE), available on Github, is regularly updated with new features and can be installed via source, RPM, or DEB.