Features

Low latency and high throughput

Bigtable is a key-value and wide-column store, ideal for fast access to structured, semi-structured, or unstructured data. This makes latency-sensitive workloads like personalization a perfect fit for Bigtable. However its distributed counters, high read and write throughput per dollar makes it also a great fit for clickstream and IoT use cases, and even batch analytics for high-performance computing (HPC) applications, including training ML models.

Write and read scalability with no limits

Bigtable decouples compute resources from data storage, which makes it possible to transparently adjust processing resources. Each additional node can process reads and writes equally well, providing effortless horizontal scalability. Bigtable optimizes performance by automatically scaling resources to adapt to server traffic, handling the sharding, replication, and query processing.

Data model flexibility

Bigtable lets your data model evolve organically. Store anything from scalars, JSON, Protocol Buffers, Avro, Arrow to embeddings, images, and dynamically add/remove new columns as needed. Deliver low-latency serving or high-performance batch analytics over raw, unstructured data in a single database.

From a single zone up to eight regions at once

Apps backed by Bigtable can deliver low-latency reads and writes with globally distributed multi-primary configurations, no matter where your users may be. Zonal instances are great for cost savings and can be seamlessly scaled up to multi-region deployments with automatic replication. When running a multi-region instance, your database is protected against a regional failure and offers industry-leading 99.999% availability.

Easy migration from NoSQL databases

Live migrations enable faster and simpler onboarding by ensuring accurate data migration with reduced effort. HBase Bigtable replication library allows for no-downtime live migrations with import and validation tools to easily load HBase snapshots into Bigtable while Dataflow templates simplify migrations from Cassandra to Bigtable.

High-performance, workload-isolated data processing

Bigtable Data Boost enables users to run analytical queries, batch ETL processes, train ML models or export data faster without affecting transactional workloads. Data Boost does not require capacity planning or management. It allows directly querying data stored in Google’s distributed storage system, Colossus using on-demand capacity letting users easily handle mixed workloads and share data worry-free.

Rich application and tool support

Easily connect to the open source ecosystem with the Apache HBase API. Build data-driven applications faster with seamless integrations with Apache Spark, Hadoop, GKE, Dataflow, Dataproc, Vertex AI Vector Search, and BigQuery. Meet development teams where they are with SQL and client libraries for Java, Go, Python, C#, Node.js, PHP, Ruby, C++, HBase and integration with LangChain.

No hidden costs

No IOPS charges, no cost for taking or restoring backups, no disproportionate read/write pricing to impact your budget as your workloads evolve.

Automated maintenance

Reduce operational costs and improve reliability for any database size. Replication and maintenance are automatic and built-in with zero downtime.

Real-time change data capture and eventing

Use Bigtable change streams to capture change data from Bigtable databases and integrate it with other systems for analytics, event triggering, and compliance.

Enterprise-grade security and controls

Customer-managed encryption keys (CMEK) with Cloud External Key Manager support, IAM integration for access and controls, support for VPC-SC, Access Transparency, Access Approval and comprehensive audit logging help ensure your data is protected and complies with regulations. Fine-grained access control lets you authorize access at table, column, or row level.

Observability

Monitor performance of Bigtable databases with server-side metrics. Analyze usage patterns with Key Visualizer interactive monitoring tool. Use query stats, table stats, and the hot tablets tool for troubleshooting query performance issues and quickly diagnose latency issues with client-side monitoring.

Disaster recovery

Take instant, incremental backups of your database cost-effectively and restore on demand. Store backups in different regions for additional resilience, easily restore between instances, or projects for test versus production scenarios.

Vertex AI Vector Search integration

Use the Bigtable to Vertex AI Vector Search template to index data in your Bigtable database with Vertex AI to perform a similarity search over vector embeddings with Vertex AI Vector Search.

LangChain integration

Easily build generative AI applications that are more accurate, transparent, and reliable with built-in kNN nearest neighbor vector search (in preview) and LangChain integration. Visit the GitHub repository to learn more.

How It Works

Bigtable instances provide compute and storage in one or more regions. Each Bigtable cluster can receive both reads and writes. Data is automatically "split" for scalability and replicated between clusters asynchronously. A distributed clock called TrueTime guarantees transactions are correctly ordered. 

Bigtable Architecture

Common Uses

AdTech and retail

Personalize experiences in real time

Track customer behavior and preferences for personalized ads, news feeds, discount offers, and product or content recommendations. Ingest high volume event streams and serve recommendations at low latency using a single database that automatically scales and rebalances for best performance. Bring data closer to your customers for best latencies with multi-region, multi-primary deployments, and reduce risk and downtime with 99.999% availability and zero maintenance.
AdTech and Retail Architecture Reference Diagram

Personalize experiences in real time

Track customer behavior and preferences for personalized ads, news feeds, discount offers, and product or content recommendations. Ingest high volume event streams and serve recommendations at low latency using a single database that automatically scales and rebalances for best performance. Bring data closer to your customers for best latencies with multi-region, multi-primary deployments, and reduce risk and downtime with 99.999% availability and zero maintenance.
AdTech and Retail Architecture Reference Diagram

Data fabric and operational analytics

Consolidate data silos and scale out legacy systems

Ingest and integrate data from multiple databases, streaming sources, and mainframes in bulk or real time using integrations with BigQuery, Dataflow, Cloud Composer, and Cloud Data Fusion to build customer data platforms, operational data stores, digital integration hubs, semantic layers, or data fabrics to support low-latency API access and scalable in-app reporting.
Data Fabric and Operational Analytics Architecture Reference Diagram

Consolidate data silos and scale out legacy systems

Ingest and integrate data from multiple databases, streaming sources, and mainframes in bulk or real time using integrations with BigQuery, Dataflow, Cloud Composer, and Cloud Data Fusion to build customer data platforms, operational data stores, digital integration hubs, semantic layers, or data fabrics to support low-latency API access and scalable in-app reporting.
Data Fabric and Operational Analytics Architecture Reference Diagram

Cybersecurity

Detect malware, payment fraud, stop spam and scams

Capture fraud signals like user activity, catalog files, malware signatures and blocklists, unstructured content like product listings and reviews to identify counterfeit goods, spam accounts, scams, compromised hardware, and fraud in real time.
Cybersecurity Architecture Reference Diagram

Detect malware, payment fraud, stop spam and scams

Capture fraud signals like user activity, catalog files, malware signatures and blocklists, unstructured content like product listings and reviews to identify counterfeit goods, spam accounts, scams, compromised hardware, and fraud in real time.
Cybersecurity Architecture Reference Diagram

Media

Deliver media content and engagement analytics

Manage playlists, augmented reality (AR) assets, book, audio, or video catalogs, watch histories, ratings and comments, track viewing progress, and serve personalized content feeds and analytics for content creators and advertisers.
Media Architecture Reference Diagram

Deliver media content and engagement analytics

Manage playlists, augmented reality (AR) assets, book, audio, or video catalogs, watch histories, ratings and comments, track viewing progress, and serve personalized content feeds and analytics for content creators and advertisers.
Media Architecture Reference Diagram

Time series and IoT

Manage time series data at any scale

From financial time series to smart homes, weather sensors, online gaming logs, factory floor telemetry, connected cars, or event sourcing architectures, ingest large amounts of data without disrupting low-latency serving workloads to support real-time reporting, alerting, and predictive maintenance. Simplify data management with TTL rules, retain data cost-effectively using the storage medium of your choice at industry leading physical storage prices, and achieve high scan throughput for batch analytics without breaking a sweat.
Time Series and IoT Architecture Reference Diagram

Manage time series data at any scale

From financial time series to smart homes, weather sensors, online gaming logs, factory floor telemetry, connected cars, or event sourcing architectures, ingest large amounts of data without disrupting low-latency serving workloads to support real-time reporting, alerting, and predictive maintenance. Simplify data management with TTL rules, retain data cost-effectively using the storage medium of your choice at industry leading physical storage prices, and achieve high scan throughput for batch analytics without breaking a sweat.
Time Series and IoT Architecture Reference Diagram

Machine learning infrastructure

Scale model training and serving

Build feature stores to support low-latency predictions, cache data from GCS for fast access by HPC clusters and ML frameworks, and snapshot model weights during training with high-throughput, low-latency reads and writes, granular access control, and workload isolation.
ML Infrastructure Architecture Reference Diagram

Learn how to use Bigtable with popular open-source feature stores.

Scale model training and serving

Build feature stores to support low-latency predictions, cache data from GCS for fast access by HPC clusters and ML frameworks, and snapshot model weights during training with high-throughput, low-latency reads and writes, granular access control, and workload isolation.
ML Infrastructure Architecture Reference Diagram

Learn how to use Bigtable with popular open-source feature stores.

Pricing

How Bigtable pricing worksBigtable pricing is based on compute capacity, database storage, backup storage, and network usage. Committed use discounts reduce the price further.
ServiceDescriptionPrice

Compute capacity

Compute capacity is provisioned as nodes.

Starting at

$0.65

per node per hour

Data storage

SSD

Pricing is based on the physical size of tables. Each replica is billed separately. Recommended for low-latency serving.

Starting at

$0.17

per GB per month

HDD

Pricing is based on the physical size of tables. Each replica is billed separately.

Starting at

$0.026

per GB per month

Backups

Pricing is based on the physical size of backups. Bigtable backups are incremental.

Starting at

$0.026

per GB per month

Network

Ingress

Free

Egress within same region

Free

Egress between regions

Starting at

$0.10

per GB

Replication

Within same region

Free

Between regions

Starting at

$0.01

per GB

Learn more about Bigtable pricing and committed use discounts.

How Bigtable pricing works

Bigtable pricing is based on compute capacity, database storage, backup storage, and network usage. Committed use discounts reduce the price further.

Compute capacity

Description

Compute capacity is provisioned as nodes.

Price

Starting at

$0.65

per node per hour

Data storage

Description

SSD

Pricing is based on the physical size of tables. Each replica is billed separately. Recommended for low-latency serving.

Price

Starting at

$0.17

per GB per month

HDD

Pricing is based on the physical size of tables. Each replica is billed separately.

Description

Starting at

$0.026

per GB per month

Backups

Description

Pricing is based on the physical size of backups. Bigtable backups are incremental.

Price

Starting at

$0.026

per GB per month

Network

Description

Ingress

Price

Free

Egress within same region

Description

Free

Egress between regions

Description

Starting at

$0.10

per GB

Replication

Description

Within same region

Price

Free

Between regions

Description

Starting at

$0.01

per GB

Learn more about Bigtable pricing and committed use discounts.

PRICING CALCULATOR

Estimate your monthly Bigtable costs, including region specific pricing and fees.

CUSTOM QUOTE

Connect with our sales team to get a custom quote for your organization.

Start your Bigtable proof of concept

User you $300 credit (new users)

Learn how to use Bigtable

Federate queries from BigQuery into Bigtable

Migrate from HBase, Cassandra, Aerospike, or DynamoDB to Bigtable

Dive into coding with examples

Business Case

Explore how other businesses built innovative apps to deliver great customer experiences, cut costs, and increase ROI with Bigtable


Explore how Box modernized their NoSQL databases with Bigtable

Box enhanced scalability and availability while reducing cost to manage, through a seamless migration.

Watch the video

Benefits and customers

Grow your business with innovative applications that scale limitlessly to meet any demand.

Get best-in-class price-performance and pay for what you use.

Migrate easily from other NoSQL databases and run hybrid or multicloud deployments with open source APIs and migration tools.

  • Equifax logo
  • PayPal logo
  • Credit Karma logo
  • Major League Baseball logo
  • The Home Depot logo
  • Fastly logo
  • fullstory logo
  • Televisa Univision logo
  • Mercadolibre logo
  • Vimeo logo
  • Evernote logo
  • Bit.ly logo
  • Squarespace logo
  • LiveRamp logo
  • OpenX logo

Partners & Integration

Take advantage of partners with Bigtable expertise to help you at every step of the journey, from assessments and business cases to migrations and building new apps on Bigtable.
  • SADA logo
  • DoIT logo
  • Searce logo
  • 66 degrees logo
  • Carahsoft logo
  • Devoteam logo
  • Cloud Ace logo
  • CloudMile logo
  • Quantiphi logo
  • Bespin Global logo
  • Huware Srl logo
  • Onix logo
  • WebEye logo
  • SoftChoice logo
  • Appsbroker logo
  • OchK logo
  • Santo Digital logo
  • MasterConcept logo
  • Polymeric Cloud logo
  • Persistent Systems logo
  • BrioTech logo
  • Baidao logo
  • Noovle logo
  • zencore logo
  • Public Cloud Group logo
  • Tangerine logo
  • Xebia logo
  • Crayon logo
  • G-gen logo
  • Comm-it logo
  • Cleardata logo
  • Digicloud Africa logo
  • IpNet logo
  • Epam logo
  • Pythian logo
  • mavenwave logo
  • Rackspace logo
  • Accenture logo
  • Megazone Soft logo
  • SADA logo
  • DoIT logo
  • Searce logo
  • 66 degrees logo
  • Carahsoft logo
  • Devoteam logo
  • Cloud Ace logo
  • CloudMile logo
  • Quantiphi logo
  • Bespin Global logo
  • Huware Srl logo
  • Onix logo
  • WebEye logo
  • SoftChoice logo
  • Appsbroker logo
  • OchK logo
  • Santo Digital logo
  • MasterConcept logo
  • Polymeric Cloud logo
  • Persistent Systems logo
  • BrioTech logo
  • Baidao logo
  • Noovle logo
  • zencore logo
  • Public Cloud Group logo
  • Tangerine logo
  • Xebia logo
  • Crayon logo
  • G-gen logo
  • Comm-it logo
  • Cleardata logo
  • Digicloud Africa logo
  • IpNet logo
  • Epam logo
  • Pythian logo
  • mavenwave logo
  • Rackspace logo
  • Accenture logo
  • Megazone Soft logo

Want to get more details about which partner or third-party integration is best for your business? Go to the partner directory.

FAQ

What type of database is Bigtable?

Bigtable is a NoSQL database service, specifically a key-value store that allows for very wide tables with tens of thousands of columns, hence also referred to as a wide-column database or a distributed multi-dimensional map. It is a NoSQL database in the "Not only SQL" sense, rather than "zero SQL".

Bigtable is most similar to popular open source projects it inspired, such as Apache HBase and Cassandra, and hence is the most common destination for customers that deal with large data volumes looking for a high-performance, cost-effective, fully managed NoSQL database solution on Google Cloud.

In addition to its Key-Value APIs, Bigtable also supports SQL queries in three different ways:

  • For low-latency application development, Bigtable offers a SQL query API that builds upon GoogleSQL with extensions for the wide-column data model resembling Cassandra Query Language (CQL).
  • For data science use cases or other kinds of batch processing and ETL, Bigtable supports SparkSQL using its Spark client.
  • For users who want to do post-hoc exploratory analysis or blend data from multiple sources for batch analytics, Bigtable data can also be accessed from BigQuery. Simply register your Bigtable tables in BigQuery and query like any other BigQuery table without any ETL or data duplication.

Bigtable offers migration tooling that enables faster and simpler onboarding by ensuring accurate data migration with reduced effort. HBase Bigtable replication library allows for no-downtime live migrations with import and validation tools to easily load HBase snapshots into Bigtable while Dataflow templates simplify migrations from Cassandra to Bigtable.

Bigtable storage is billed per GB used similar to a serverless model. Bigtable also offers linear horizontal scaling and can automatically scale up and down compute resources in response to demand fluctuations. Hence it doesn’t require a long term capacity commitment for storage or compute. However, pricing for low-latency compute is capacity-based and billed per node, not per request where each node can serve up to 17K requests per second. This makes Bigtable price more favorable for larger workloads but less ideal for small applications, which may be more suitable for Google Cloud databases such as Firestore.

For batch data processing Bigtable offers Data Boost, which bills in Serverless Processing Units (SPU).

Learn more
Google Cloud
  • ‪English‬
  • ‪Deutsch‬
  • ‪Español‬
  • ‪Español (Latinoamérica)‬
  • ‪Français‬
  • ‪Indonesia‬
  • ‪Italiano‬
  • ‪Português (Brasil)‬
  • ‪简体中文‬
  • ‪繁體中文‬
  • ‪日本語‬
  • ‪한국어‬
Console
Google Cloud