Provisioned nodes

Bigtable keeps track of how many nodes exist in your instance's clusters during each hour. You are charged each hour for the maximum number of nodes that exist during that hour, multiplied by the hourly rate. Bigtable bills a minimum of one hour for each node you provision.

Node charges are for provisioned resources, regardless of node usage. Charges apply even if your cluster is inactive.

Data Boost serverless compute

Data Boost for Bigtable is an on-demand, serverless compute option. It's designed to isolate throughput-intensive read use cases like pipeline jobs and queries from your provisioned node compute resources. To use Data Boost, you need at least one active Bigtable cluster with at least one provisioned node.

With Data Boost for Bigtable, you pay for the compute resources that are necessary to process your job or query. Bigtable automatically allocates those resources when you need them.

Bigtable Data Boost usage is measured in serverless processing units or SPUs. One SPU is equivalent to the processing power of approximately 1/1000th of a node.

You are charged for the SPU-seconds that you consume, with a minimum of at least 60 SPU-seconds per request.

No charges apply during periods without any Data Boost usage. No additional storage or networking fees apply.

Committed use discounts are not available for Bigtable Data Boost.

Bigtable frequently measures the average amount of data in your Bigtable tables during a short time interval. For billing purposes, these measurements are combined into an average over a one-month period, and this average is multiplied by the monthly rate.

You are billed for storage based on the physical size of your tables and backups. Bigtable compresses your data, reducing the amount of storage it occupies. In addition, Bigtable regularly goes through a process called compaction to further reduce your storage cost. Compaction typically happens within a week of data being written or deleted.

Deleted data becomes inaccessible immediately; however, you are charged for storage of the data until Bigtable compacts the table. In addition, if you store multiple versions of a value in a column and set an expiration time for older versions or limit the number of versions, you are also charged to store the expired values prior to garbage collection, which follows a similar cadence to compaction.

For instances that contain multiple clusters, Bigtable keeps a separate copy of your data with every cluster, and you are charged for every copy of your data.

You are billed only for the backup storage that you use, and a backup can never be larger than the original table. A backup is billed from the time it is created until the time it is deleted.

Standard backups

Because of Bigtable's built-in storage optimizations, the cost of the backup might sometimes be less than the cost of a full physical copy of the table backup.

Hot backups

Hot backups are production-ready backups that are optimized for speedy recovery. A hot backup consumes the same amount of storage as the source table at the time the backup is created, and its size does not change over time.

Replication rates

North America

Europe

Asia Pacific

Oceania/Indonesia

South America

Singapore

Japan

Some types of data transfer are subject to bandwidth charges, as shown below.

Data transfer in: Free

Intra-zone data transfer: Free

Inter-zone data transfer in the same region: Free

Inter-region data transfer in the same continent (per GiB): $0.01

Inter-continental data transfer: At outbound internet data transfer rates

If you pay in a currency other than USD, the prices listed in your currency on Cloud Platform SKUs apply.

Source region: North America, Europe, Asia Pacific or South America

Source region: Oceania/Indonesia

Source region: Singapore

Source region: Japan

One-time, initial replication traffic

Each instance's clusters are located in zones, and each zone is located in a region. If you add a new cluster in a region, and none of the instance's existing clusters are in that region, you are charged a one-time cost for the initial data copy to the new cluster. For this initial copy, you are charged the network rate for replication to the existing cluster that is geographically closest to the new cluster, multiplied by the amount of data in your tables. There is no charge for replication between clusters in a single region.

If you restore a new table from a backup, and the new table is in an instance that uses replication, you are charged the standard network rate for replication for the initial data copy to every cluster in the instance.

To illustrate the cost of adding a cluster to an instance, let's say your instance has 9 TiB of data, and its clusters are in us-east1 (South Carolina) and europe-west1 (Belgium). If you add a new cluster to the instance in the us-central1 (Iowa) region, your data is copied to the new cluster from the existing cluster in us-east1, which is the closest cluster in the instance. You are charged a rate of $0.01 per GiB, the rate for network traffic within the US for replication. As a result, the initial replication cost is $92.16:

9 TiB * 1,024 GiB/TiB * $0.01 per GiB = $92.16

In another example, let's say your instance has 9 TiB of data, and its clusters are in us-east4 (Northern Virginia) and us-west2(Los Angeles). If you add a new cluster in northamerica-northeast2 (Toronto), your data is copied to the new cluster from the existing cluster in us-east4, which is the closest cluster in the instance. You are charged a rate of $0.12 per GiB, the rate for replication of 0 GiB to 1,024 GiB within North America (not the same as the rate for replication within the US). In this example, the initial replication cost is $1,105.92:

9 TiB * 1,024 GiB/TiB * $0.12 per GiB = $1,105.92

Ongoing replication traffic

When you write to a cluster, Bigtable automatically replicates the write from that cluster (the "source cluster") to all of the other clusters in the instance (the "destination clusters"). If a destination cluster is in a different region than the source cluster, you are charged to replicate the write to the destination cluster. The charge is equal to the network rate for replication to the destination cluster, multiplied by the amount of data that you wrote to the source cluster. You incur a separate charge for every destination cluster that is in a different region than the source cluster.

For example, let's say you have a 3-cluster instance with clusters in us-east1 (South Carolina), europe-west1 (Belgium), and australia-southeast1 (Sydney). Over the course of a month, if your application writes 5 TiB of data to the cluster in us-east1, that 5 TiB of data is replicated from us-east1 to europe-west1 at the rate of $0.11 per GiB, which is the cost for replication of 1 TiB to 10 TiB from the US to Europe. The data is also replicated from us-east1 to australia-southeast1 at the rate of $0.18 per GiB, which is the cost for replication of 1 TiB to 10 TiB from the US to Australia. In this example, the 30-day replication cost for the 5 TiB of data originally written to the cluster in us-east1 is $1,484.80:

(5 TiB * 1,024 GiB/TiB * $0.11 per GiB) + (5 TiB * 1,024 GiB/TiB * $0.18 per GiB) = $1,484.80

Suppose you have a Bigtable instance that has one cluster. The cluster has one node, and your application server is in the same region as Bigtable. During a 30-day billing cycle, you use the following Bigtable resources:

• 1 instance in us-central1 (Iowa) with a single cluster that has 1 node throughout the month
• Average of 50 GiB of data stored on SSD drives in us-central1
• No data transfer

Assuming that you are billed in USD, your Bigtable bill for the billing cycle is broken down as follows. Additional fees would apply for any other Google Cloud services used during the billing cycle, such as Compute Engine instances.

Bigtable nodes

• 1 cluster * 1 node * 30 days * 24 hours/day * $0.65 per node per hour in us-central1: $468.00

Storage

• 1 cluster * 50 GiB * $0.17 per GiB in us-central1: $8.50

Network

• No data transfer

Monthly total

In this example, the total monthly bill for Bigtable is $476.50.

Suppose you have a Bigtable instance that has 1 cluster, with no replication. During a 30-day billing cycle, you use the following Bigtable resources:

• 1 instance in us-central1 (Iowa) with a single cluster, which has 18 nodes during days 1-10 and 25 nodes during days 11-30
• Average of 30 TiB of data stored on SSD drives in us-central1
• 10 TiB of data transfer into us-central1
• 50 GiB of data transfer out to us-central1
• 25 GiB of data transfer out to europe-west1

Assuming that you are billed in USD, your Bigtable bill for the billing cycle is broken down as follows. Additional fees would apply for any other Google Cloud services used during the billing cycle, such as Compute Engine instances.

Bigtable nodes

Days 1-10

• 1 cluster * 18 nodes * 10 days * 24 hours/day * $0.65 per node per hour in us-central1: $2,808.00

Days 11-30

• 1 cluster * 25 nodes * 20 days * 24 hours/day * $0.65 per node per hour in us-central1: $7,800.00

Storage

• 1 cluster * 30 TiB * 1,024 GiB/TiB * $0.17 per GiB in us-central1: $5,222.40

Network

• 10 TiB of data transfer into us-central1: No charge
• 50 GiB of data transfer out to us-central1: No charge
• 25 GiB of data transfer out to europe-north1 * $0.12 per GiB: $3.00

Monthly total

In this example, the total monthly bill for Bigtable is $15,833.40.

Suppose you have a Bigtable instance that has 2 clusters, which means that replication is enabled. The clusters are in different zones in the same region. During a 30-day billing cycle, you use the following Bigtable resources:

• 1 instance in us-central1 (Iowa) with 2 clusters; each cluster has 18 nodes during days 1-10 and 25 nodes during days 11-30
• Average of 30 TiB of data stored on SSD drives for each cluster
• 10 TiB of data transfer into us-central1
• 50 GiB of data transfer out to us-central1
• 25 GiB of network egress to europe-north1

Note: This example is similar to the example single cluster with multiple nodes, except the instance in this example has 2 clusters in a single region.

Assuming that you are billed in USD, your Bigtable bill for the billing cycle is broken down as follows. Additional fees would apply for any other Google Cloud services used during the billing cycle, such as Compute Engine instances.

Bigtable nodes

Days 1-10

• 2 clusters * 18 nodes * 10 days * 24 hours/day * $0.65 per node per hour in us-central1: $5,616.00

Days 11-30

• 2 clusters * 25 nodes * 20 days * 24 hours/day * $0.65 per node per hour in us-central1: $15,600.00

Storage

• 2 clusters * 30 TiB * 1,024 GiB/TiB * $0.17 per GiB in us-central1: $10,444.80

Network

• 10 TiB of data transfer into us-central1: No charge
• 50 GiB of data transfer out to us-central1: No charge
• 25 GiB of data transfer out to europe-north1 * $0.12 per GiB: $3.00

Monthly total

In this example, the total monthly bill for Bigtable is $31,663.80.

Suppose you have a Bigtable instance that has 2 clusters, each in a different Bigtable region. During a 30-day billing cycle, you use the following Bigtable resources:

• 1 instance with 1 cluster in us-central1 (Iowa) and 1 cluster in asia-south1 (Mumbai); each cluster has 18 nodes during days 1-10 and 25 nodes during days 11-30
• Average of 30 TiB of data stored on SSD drives for each cluster
• 10 TiB of network data transfer into us-central1
• 2 TiB of network ingress to asia-south1
• 50 GiB of network data transfer out to us-central1
• 25 GiB of network egress to europe-north1
• 10 TiB of writes replicated from us-central1 to asia-south1
• 2 TiB of writes replicated from asia-south1 to us-central1

Note: This example is similar to the example 2 clusters in the same region, except the clusters in this example are in different regions.

Assuming that you are billed in USD, your Bigtable bill for the billing cycle is broken down as follows. Additional fees would apply for any other Google Cloud services used during the billing cycle, such as Compute Engine instances.

Bigtable nodes

Days 1-10

• 1 cluster * 18 nodes * 10 days * 24 hours/day * $0.65 per node per hour in us-central1: $2,808.00
• 1 cluster * 18 nodes * 10 days * 24 hours/day * $0.748 per node per hour in asia-south1: $3,231.36

Days 11-30

• 1 clusters * 25 nodes * 20 days * 24 hours/day * $0.65 per node per hour in us-central1: $7,800.00
• 1 clusters * 25 nodes * 20 days * 24 hours/day * $0.748 per node per hour in asia-south1: $8,976.00

Storage

• 1 cluster * 30 TiB * 1,024 GiB/TiB * $0.17 per GiB in us-central1: $5,222.40
• 1 cluster * 30 TiB * 1,024 GiB/TiB * $0.196 per GiB in asia-south1: $6,021.12

Network

• 10 TiB of ingress to us-central1: No charge
• 2 TiB of ingress to asia-south1: No charge
• 50 GiB of data transfer out to us-central1: No charge
• 25 GiB of data transfer out to europe-north1 * $0.12 per GiB: $3.00

Replication

• 10 TiB of writes replicated from us-central1 to asia-south1 * 1,024 GiB/TiB * $0.12 per GiB: $1,228.80
• 2 TiB of writes replicated from asia-south1 to us-central1 * 1,024 GiB/TiB * $0.12 per GiB: $245.76

Monthly total

In this example, the total monthly bill for Bigtable is $35,536.44.

Suppose you have a Bigtable instance that has 3 clusters, each in a different Bigtable region. During a 30-day billing cycle, you use the following Bigtable resources:

• 1 instance with 1 cluster in us-central1 (Iowa), 1 cluster in asia-south1 (Mumbai), and 1 cluster in europe-west2 (London); each cluster has 15 nodes on every day of the 30-day month
• Average of 50 TiB of data stored on SSD drives for each cluster
• 10 TiB of writes to each cluster; writes are then replicated to the other 2 clusters
• 25 GiB of data transfer out to asia-northeast1

Assuming that you are billed in USD, your Bigtable bill for the billing cycle is broken down as follows. Additional fees would apply for any other Google Cloud services used during the billing cycle, such as Compute Engine instances.

Bigtable nodes

• 1 cluster * 15 nodes * 30 days * 24 hours/day * $0.65 per node per hour in us-central1: $7,020.00
• 1 cluster * 15 nodes * 30 days * 24 hours/day * $0.748 per node per hour in asia-south1: $8,078.40
• 1 cluster * 15 nodes * 30 days * 24 hours/day * $0.78 per node per hour in europe-west2: $8,424.00

Storage

• 1 cluster * 50 TiB * 1,024 GiB/TiB * $0.17 per GiB in us-central1: $8,704.00
• 1 cluster * 50 TiB * 1,024 GiB/TiB * $0.196 per GiB in asia-south1: $10,035.20
• 1 cluster * 50 TiB * 1,024 GiB/TiB * $0.20 per GiB in europe-west2: $10,240.00

Network

• 10 TiB of data transfer into us-central1: No charge
• 10 TiB of data transfer into asia-south1: No charge
• 10 TiB of data transfer into europe-west2: No charge
• 50 GiB of data transfer into us-central1: No charge
• 25 GiB of data transfer into asia-northeast1 * $0.12 per GiB: $3.00

Replication

• 10 TiB of writes replicated from us-central1 to asia-south1 * 1,024 GiB/TiB * $0.11 per GiB: $1,126.40
• 10 TiB of writes replicated from us-central1 to europe-west2 * 1,024 GiB/TiB * $0.11 per GiB: $1,126.40
• 10 TiB of writes replicated from asia-south1 to us-central1 * 1,024 GiB/TiB * $0.11 per GiB: $1,126.40
• 10 TiB of writes replicated from asia-south1 to europe-west2 * 1,024 GiB/TiB * $0.11 per GiB: $1,126.40
• 10 TiB of writes replicated from europe-west2 to us-central1 * 1,024 GiB/TiB * $0.11 per GiB: $1,126.40
• 10 TiB of writes replicated from europe-west2 to asia-south1 * 1,024 GiB/TiB * $0.11 per GiB: $1,126.40

Monthly total

In this example, the total monthly bill for Bigtable is $59,263.00.