Frequently asked questions about Azure Cosmos DB for Table

There are some behavior differences that users coming from Azure Table storage who want to create tables with the Azure Cosmos DB for Table should be aware of:

• Azure Cosmos DB for Table uses a reserved capacity model in order to ensure guaranteed performance but this means that one pays for the capacity as soon as the table is created, even if the capacity isn't being used. With Azure Table storage one only pays for capacity that's used. This helps to explain why API for Table can offer a 10 ms read and 15 ms write SLA at the 99th percentile while Azure Table storage offers a 10-second SLA. But as a consequence, with API for Table tables, even empty tables without any requests, cost money in order to ensure the capacity is available to handle any requests to them at the SLA offered by Azure Cosmos DB.

• Query results returned by the API for Table aren't sorted in partition key/row key order as they are in Azure Table storage.

• Row keys can only be up to 255 bytes.

• Batches can only have up to 2 MBs.

• CORS isn't currently supported.

• Table names in Azure Table storage aren't case-sensitive, but they are in Azure Cosmos DB for Table.

• Some of Azure Cosmos DB's internal formats for encoding information, such as binary fields, are currently not as efficient as one might like. Therefore this can cause unexpected limitations on data size. For example, currently one couldn't use the full one Meg of a table entity to store binary data because the encoding increases the data's size.

• Azure Cosmos DB reserves the entity property names ID, rid, etag, and ResourceId and they're currently not supported.

• TableQuery TakeCount isn't limited to 1000.

• In terms of the REST API, Azure Table storage (but not Azure Cosmos DB for Table) supports the following endpoints/query options:

  Rest Methods	Rest Endpoint/Query Option	Doc URLs	Explanation	Supported in Table Storage	Supported in API for Table
  GET, PUT	/?Restype=service@comp=properties	Set Table Service Properties and Get Table Service Properties	This endpoint is used to set CORS rules, storage analytics configuration, and logging settings. CORS is currently not supported and analytics and logging are handled differently in Azure Cosmos DB than Azure Storage Tables	Yes	No
  OPTIONS	/<table-resource-name>	Preflight CORS table request	This is part of CORS which Azure Cosmos DB doesn't currently support.	Yes	No
  GET	/?Restype=service@comp=stats	Get Table Service Stats	Provides information how quickly data is replicating between primary and secondaries. This isn't needed in Azure Cosmos DB as the replication is part of writes.	Yes	No
  GET, PUT	/mytable?comp=acl	Get Table ACL and Set Table ACL	This gets and sets the stored access policies used to manage Shared Access Signatures (SAS).	Yes	No

• Azure Cosmos DB for Table only supports the JSON format, not ATOM.

• For the .NET SDK in particular, there are some classes and methods that Azure Cosmos DB doesn't currently support.

  • CloudTableClient class
    • \ServiceProperties
    • \ServiceStats
  • CloudTable class
    • SetPermissions
    • GetPermissions

No, existing storage SDKs should still work. However, it's recommended that one always gets the latest SDKs for the best support and in many cases superior performance. See the list of available languages in the Introduction to Azure Cosmos DB for Table.

The connection string is:

DefaultEndpointsProtocol=https;AccountName=<AccountNamefromCosmosDB;AccountKey=<FromKeysPaneofCosmosDB>;TableEndpoint=https://<AccountName>.table.cosmosdb.azure.cn

You can get the connection string from the Connection String page in the Azure portal.

Some settings are handled on the CreateCloudTableClient method and other via the app.config in the appSettings section in the client application. For information about config settings, see Azure Cosmos DB capabilities.

None. There are no changes for existing or new customers who are using the existing Azure Table storage SDKs.

You can use the Azure portal to browse the data. You can also use the API for Table code or the tools mentioned in the next answer.

You can use the Azure Storage Explorer.

Tools with the flexibility to take a connection string in the format specified previously can support the new API for Table. A list of table tools is provided on the Azure Storage Client Tools page.

Yes, optimistic concurrency is provided via the use of the ETag mechanism.

Yes, the API for Table supports OData query and LINQ query.

Yes, you can connect by creating two separate instances of the CloudTableClient, each pointing to its own URI via the connection string.

AzCopy is supported.

Azure Cosmos DB is designed to provide unlimited storage via the use of horizontal scaling. The service can monitor and effectively increase your storage.

You can use the API for Table Metrics pane to monitor requests and storage usage.

You can use the capacity estimator to calculate the TableThroughput that's required for the operations. For more information, see Estimate Request Units and Data Storage. In general, you can show your entity as JSON and provide the numbers for your operations.

Not at this time.

Yes, the same API is supported.

No, you can create and use existing Azure Table storage assets without interruption of any kind. However, if you don't use the API for Table, you can't benefit from the automatic index, the extra consistency option, or multiple-regional distribution.

You can use the Azure Cosmos DB portal's multiple-region replication settings to add regions that are suitable for your application. To develop a multiple-regionally distributed application, you should also add your application with the PreferredLocation information set to the local region for providing low read latency.

You can use the Azure Cosmos DB multiple-region replication portal pane to add a region and then fail over to the required region. For instructions, see Developing with multi-region Azure Cosmos DB accounts.

To help read from the local location, use the PreferredLocation key in the app.config file. For existing applications, the API for Table throws an error if LocationMode is set. Remove that code, because the API for Table picks up this information from the app.config file.

Azure Cosmos DB provides well-reasoned trade-offs between consistency, availability, and latency. Azure Cosmos DB offers five consistency levels to API for Table developers, so you can choose the right consistency model at the table level and make individual requests while querying the data. When a client connects, it can specify a consistency level. You can change the level via the consistencyLevel argument of CreateCloudTableClient.

The API for Table provides low-latency reads with "Read your own writes," with Bounded-staleness consistency as the default. For more information, see Consistency levels.

By default, Azure Table storage provides Strong consistency within a region and Eventual consistency in the secondary locations.

Yes, for information about how to benefit from the distributed nature of Azure Cosmos DB, see Consistency levels. Because guarantees are provided for the consistency levels, you can use them with confidence.

Azure Cosmos DB commits the data durably in the local region and pushes the data to other regions immediately in a matter of milliseconds. This replication is dependent only on the round-trip time (RTT) of the datacenter. To learn more about the global-distribution capability of Azure Cosmos DB, see Azure Cosmos DB: A multiple-regionally distributed database service on Azure.

With Azure Cosmos DB, you can set the consistency level at the container level (on the table). By using the .NET SDK, you can change the level by providing the value for TableConsistencyLevel key in the app.config file. The possible values are: Strong, Bounded Staleness, Session, Consistent Prefix, and Eventual. For more information, see Tunable data consistency levels in Azure Cosmos DB. The key idea is that you can't set the request consistency level at more than the setting for the table. For example, you can't set the consistency level for the table at Eventual and the request consistency level at Strong.

The API for Table uses the multiple-regionally distributed platform of Azure Cosmos DB. To ensure that your application can tolerate datacenter downtime, enable at least one more region for the account in the Azure Cosmos DB portal Developing with multi-region Azure Cosmos DB accounts. You can set the priority of the region by using the portal Developing with multi-region Azure Cosmos DB accounts.

You can add as many regions as you want for the account and control where it can fail over to by providing a failover priority. To use the database, you need to provide an application there too. When you do so, your customers won't experience downtime. The latest .NET client SDK is auto homing but the other SDKs aren't. That is, it can detect the region that's down and automatically fail over to the new region.

Yes, the API for Table uses the platform of Azure Cosmos DB for backups. Backups are made automatically. For more information, see Online backup and restore with Azure Cosmos DB.

Yes, all attributes of an entity are indexed by default. For more information, see Azure Cosmos DB: Indexing policies.

Yes, Azure Cosmos DB for Table provides automatic indexing of all attributes without any schema definition. This automation frees developers to focus on the application rather than on index creation and management. For more information, see Azure Cosmos DB: Indexing policies.

Yes, you can change the indexing policy by providing the index definition. You need to properly encode and escape the settings.

The indexing policy can only be set in the portal at Data Explorer, navigate to the specific table you want to change and then go to the Scale & Settings->Indexing Policy, make the desired change and then Save.

The API for Table provides the same query functionality as Azure Table storage. Azure Cosmos DB also supports sorting, aggregates, geospatial query, hierarchy, and a wide range of built-in functions. For more information, see SQL queries.

You should change TableThroughput when either of the following conditions applies:

• You're performing an extract, transform, and load (ETL) of data, or you want to upload numerous data in short amount of time.
• You need more throughput from the container or from a set of containers at the back end. For example, you see that the used throughput is more than the provisioned throughput, and you're getting throttled. For more information, see Set throughput for Azure Cosmos DB containers.

Yes, you can use the Azure Cosmos DB portal's scale pane to scale the throughput. For more information, see Set throughput.

Yes, if you don't override the TableThroughput via app.config and don't use a precreated container in Azure Cosmos DB, the service creates a table with throughput of 400.

None. There's no change in price for existing Azure Table storage customers.

The price depends on the allocated TableThroughput.

If the request rate is more than the capacity of the provisioned throughput for the underlying container or a set of containers, you get an error, and the SDK retries the call by applying the retry policy.

Azure Cosmos DB sets a default throughput for your container if you don't provide one in the app.config file or via the portal.

Azure Cosmos DB provides guarantees for performance and latency, with upper bounds on operation. This guarantee is possible when the engine can enforce governance on the tenant's operations. Setting TableThroughput ensures that you get the guaranteed throughput and latency, because the platform reserves this capacity and guarantees operational success.

By using the throughput specification, you can elastically change it to benefit from the seasonality of your application, meet the throughput needs, and save costs.

Azure Cosmos DB is designed to be a multiple-regionally distributed, SLA-based system with guarantees for availability, latency, and throughput. When you reserve throughput in Azure Cosmos DB, it's guaranteed, unlike the throughput of other systems. Azure Cosmos DB provides more capabilities that customers have requested, such as secondary indexes and multiple-regional distribution.

Azure Cosmos DB is an SLA-based system that provides unlimited scale, with guarantees for latency, throughput, availability, and consistency. To ensure guaranteed premium performance, make sure that your data size and index are manageable and scalable. The 20-GB limit on the number of entities or items per partition key is to ensure that we provide great lookup and query performance. To ensure that your application scales well, even for Azure Storage, we recommend that you not create a hot partition by storing all information in one partition and querying it.

Yes. Because the surface area of the API for Table is similar to that of the Azure Table storage SDK, the partition key provides an efficient way to distribute the data. The row key is unique within that partition. The row key needs to be present and can't be null as in the standard SDK. The length of RowKey is 255 bytes and the length of PartitionKey is 1 KB.

Azure Table storage and Azure Cosmos DB for Table use the same SDKs so most of the errors are the same.

Azure Cosmos DB is an SLA-based system that provides latency, throughput, availability, and consistency guarantees. Because it's a provisioned system, it reserves resources to guarantee these requirements. The rapid rate of creation of tables is detected and throttled. We recommend that you look at the rate of creation of tables and lower it to less than 5 per minute. Remember that the API for Table is a provisioned system. The moment you provision it, you begin to pay for it.

You can share your feedback in any of the following ways:

• Microsoft Q&A question page
• Stack Overflow. Stack Overflow is best for programming questions. Make sure your question is on-topic and provide as many details as possible, making the question clear and answerable.

Related content

• Build a API for Table app with .NET SDK and Azure Cosmos DB
• Build a Java app to manage Azure Cosmos DB for Table data

Next step