Cloud Time Series Database Evaluation: GridDB Cloud vs. InfluxDB Cloud vs. MongoDB Atlas. Click here to read now!
GridDB Developer
Download

FAQ

GridDB Fundamentals

Query Interfaces

Data Model

Scalability & High Availability

Deployment & Cloud

Ecosystem

Others

What is GridDB?
GridDB is a highly scalable, in-memory time series database built for Big Data and IoT (Internet of Things) workloads. It is designed to ingest high-frequency, high-volume data from large numbers of sensors and devices while maintaining strong consistency and durability. GridDB is developed by Toshiba, and its Community Edition is open source.
Compared to other databases, what are GridDB’s main advantages?
GridDB’s distinguishing feature is its Key-Container data model, which maps naturally onto IoT systems: each device or data source gets its own container, and purpose-built TimeSeries containers handle timestamped data with native time-based operations. Combined with its memory-first architecture and a dual interface that offers both NoSQL (TQL) and SQL access to the same data, GridDB is well suited to real-time ingestion and analysis at scale.
What makes GridDB so fast?
GridDB uses a “memory first, storage second” architecture. “Hot” data is held in memory for fast reads and writes, while a hybrid memory-and-disk design lets a single node manage datasets far larger than its available RAM. An event-driven engine designed to minimize per-operation overhead keeps ingestion and query latency low.
What hardware and platforms does GridDB support?
GridDB runs on Linux and is distributed for common distributions. For the current, authoritative list of supported operating systems, architectures, and language clients, see the Supported Platforms section of the official documentation.
What are the differences between GridDB editions?
GridDB comes in three editions: Community Edition (CE), Enterprise Edition (EE), and GridDB Cloud. Community Edition is the free, open-source edition (AGPL v3 server, Apache 2.0 client libraries) and runs as a single node. Enterprise Edition adds multi-node clustering and high availability, security features (TLS/SSL encryption and LDAP authentication), integrated operations-management tooling (rolling upgrades, online backup, management GUI), ODBC connectivity, and commercial support. GridDB Cloud is the fully-managed service, offered through the Azure Marketplace, with the same core interfaces.
Does GridDB support transactions?
Yes. Transactions are supported on a per-container basis, meaning GridDB containers are ACID-compliant.
What is ACID?
GridDB being ACID-compliant means database transactions abide by each rule in the acronym:
Atomicity: a transaction is all-or-nothing.
Consistency: all data changes abide by the rules set by the schema and administrator.
Isolation: concurrent transactions do not interfere with one another.
Durability: once committed, data stays committed and is safe even after a failure.
Does GridDB support SQL?
Yes. GridDB provides two interfaces over the same data: a NoSQL interface, queried with TQL (GridDB’s query language for search, aggregation, and time-based operations), and a NewSQL interface, which supports standard SQL over JDBC. Both are available in the open-source Community Edition — the NewSQL/SQL interface has been part of CE since version 4.5. Because the two interfaces share the same underlying data, a container created through the NoSQL interface can be queried as a table through SQL, and a table created through SQL can be accessed as a container through the NoSQL interface. (ODBC connectivity is available in Enterprise Edition.)
What is the difference between the NoSQL and NewSQL interfaces?
The two interfaces target different use cases against the same data. The NoSQL interface is optimized for high-response, high-frequency collection and retrieval, and is queried with TQL. The NewSQL interface exposes the data as relational tables accessible with standard SQL over JDBC (and ODBC in Enterprise Edition), which is ideal for analytics, BI tools, and integrating with existing relational-database applications. A “container” in the NoSQL interface and a “table” in the NewSQL interface refer to the same object. You can use whichever interface — or both — suits a given use case.
What is TQL?
TQL is GridDB’s query language for the NoSQL interface. It provides SQL-like operations focused on search, aggregation, and time-series functions (such as sampling and interpolation) over a container. For full relational-style SQL across tables, use the NewSQL/SQL interface instead.
What is a Container?
A container is GridDB’s core data structure; you can think of it like a table in a relational database, with a defined schema of columns and rows. GridDB has two kinds: Collection containers for general-purpose data (strings, booleans, arrays, and so on) and TimeSeries containers, which are purpose-built for timestamped data and support time-specific operations. Each container has its own schema.
What is a Row?
As in the relational model, a row is a single record that conforms to the schema defined by its container. A row can have a key, but a key is not mandatory.
What time-series features does GridDB provide?
Beyond TimeSeries containers, GridDB offers time-series operations through TQL (such as time-window aggregation, sampling, and interpolation), built-in compression of time-series data to reduce storage footprint, and an expiry-based release (TTL) capability that automatically ages out data past a defined retention period.
How does GridDB scale? Does it scale horizontally?
GridDB’s architecture is designed for horizontal scale-out: in a multi-node cluster, data is automatically distributed across nodes, the master is elected autonomously so there is no single point of failure, and nodes can be added or removed online without stopping service. Multi-node clustering is a feature of GridDB Enterprise Edition (and GridDB Cloud). The open-source Community Edition runs as a single node — though, thanks to its hybrid memory-and-disk design, a single node can still manage datasets far larger than its memory.
How does GridDB handle replication and failover?
In a GridDB cluster, replicas (duplex, triplex, or higher) are maintained across nodes according to your availability requirements. Clients cache data-placement information, so when a node failure is detected, access automatically fails over to a replica and continues without data loss. Cluster replication and automatic failover are part of Enterprise Edition; the replication factor is configured in the cluster definition file (gs_cluster.json).
Can I build a cluster from commodity hardware?
Yes. GridDB Enterprise Edition is designed to scale out across clusters of commodity hardware, and capacity can be expanded online as your data grows.
Does GridDB run in a public cloud environment?
Yes. You can self-host GridDB (Community or Enterprise Edition) on any major public cloud, including Azure, AWS, and Google Cloud.
What is GridDB Cloud?
GridDB Cloud is Toshiba’s fully-managed GridDB service, available through the Azure Marketplace. It lets you use GridDB — including its NoSQL and SQL interfaces and a RESTful Web API — without provisioning or operating the underlying infrastructure yourself.
What languages and client libraries does GridDB support?
GridDB provides official client libraries for Java and C, with additional clients for Python, Go, Node.js, and PHP, as well as a RESTful Web API for language-agnostic access over HTTP. SQL access is available through the JDBC driver (and ODBC in Enterprise Edition).
What tools and platforms does GridDB integrate with?
GridDB integrates with common data-pipeline and visualization tools, including Apache Kafka (via a connector), Grafana for dashboards, and collection agents such as Telegraf, Fluentd, and Logstash. Its JDBC interface also allows it to work with standard SQL and BI tooling.
There is something not in the FAQ that I need to ask.
You can ask questions on the GridDB GitHub Discussions page or on Stack Overflow using the griddb tag. One of our engineers will be glad to help.