Using multiple regions
You can replicate data to multiple regions for high availability scenarios to ensure active-active applications failover models. Multiple regions also ensure application data availability for locality purposes with the added value of cost savings.
Having multiple regions increase may your billing. For more, see Pricing and billing.
Video introduction
See this short video introduction to the Astra DB multi-region implementation:
Eventual consistency model and multi-region updates
DataStax Astra DB follows the eventual consistency model. Depending on the selected consistency level, data written to one region might not be immediately accessible to other regions in the same database.
If you are using the EACH_QUORUM consistency level for normal updates or SERIAL consistency level for Lightweight Transactions (LWT), then data is immediately accessible on all regions provided the operation successfully completes.
These consistency levels are for only write requests.
For all other consistency levels, data might not be immediately accessible. The time span is normally within a few minutes to fully replicate the data. However, it could take longer, and possibly span one or more days. There are several contributing factors to the latter scenario, such as the workload volume, the number of regions, the data repair operations, and network resources.
For more, see the FAQs in this topic.
Serverless databases
|
Multiple regions is available on only pay as you go and annual plans. |
|
If you are adding multiple regions to your database, you can use each region only once. You cannot add the same region to the same database more than one time. |
Add a new region to your database
-
From your database’s Dashboard, select Add Region.
-
Select your desired region from the dropdown menu of available choices.
You can review your selected region and its cost below the dropdown menu.
You can add only a single region at one time.
-
Select Add Region to add the region to your database.
After you add the new region, your new region appears in the list of regions on your database dashboard.
Your existing region status will change to Maintenance, but will still be available for operations.
|
The information displayed on the Connect page for your database is region specific. |
Delete a region from your database
|
The information displayed on the Connect page for your database is region specific. |
|
Removing a region is not reversible. Proceed with caution. |
-
From you database Dashboard, select the overflow menu for the database you want to delete.
-
Select Delete.
-
Type "delete" in the text bar.
-
Select Delete to delete your region.
Limitations
If there are network issues between regions, the following operations might not be available or have limited availability:
-
Schema changes
-
Table truncation
-
Lightweight transactions
-
Data repair
While adding a new region, you cannot drop a table or keyspace and you cannot truncate a table.
Data sovereignty
Astra DB serverless replicates all data in the database to all of a database’s regions. By contrast, multiple keyspaces in Apache Cassandra® and DataStax Enterprise (DSE) allow a database to replicate some tables to a subset of regions. To achieve the same behavior as Cassandra or DSE, create a separate Astra DB instance that adheres to the necessary region restrictions. The database client will need to add a separate connection for the additional database and send queries to the appropriate connection depending on the table being queried.
Astra DB Serverless multi-region FAQ
While creating an Astra DB database, can I select multiple regions?
No. The workflow is to create a database with a selected region; then once the launched database reaches an Active status, you may add a region.
How many regions can I add?
The only limitation is the number of regions that are available in Astra DB. See Database Regions for the list of supported regions.
How many regions can I add or remove at a time?
The current implementation allows one addition or removal at a time.
How quickly are updates applied to all regions in the same database cluster?
It depends on several factors. Astra DB follows the eventual consistency model. As a result, data written to one region might not be immediately accessible in other regions in the same database cluster. The time span is normally within a few minutes to fully replicate the data. However, it could take longer, and possibly span one or more days. There are several contributing factors to the latter scenario, such as the workload volume, the number of regions, the data repair operations, and network resources.
Can I remove a region that I no longer need?
Yes, this action is supported with Astra DB Serverless. Exception: you cannot remove the original region selected when you created the Astra DB database. The original region can be deleted after all other regions are deleted by using the Terminate database operation.
Will I lose data when I remove a region?
No. All data in the datacenter is replicated prior to being decommissioned.
Can I park a region?
No. The best option, in this case, is to remove the region.
Can I specify replication rules by keyspace?
By default, replication is applied at the database level. With Astra DB Serverless, there is no upfront database setup costs. The replication goal may be addressed by separating workloads by database without an impact on costs.
Is the Access List configuration by region?
The Access List configuration applies at the database level.
How are backups managed?
-
Backups are taken on the region in which the database was first created.
-
The restore process follows this workflow: restore to the original region, and then add the other regions.
-
For help with restore, please contact DataStax Support.
What’s next?
See additional database management topics.