Azure DocumentDB pricing explained

Azure DocumentDB (with MongoDB compatibility) pricing has only two parts: compute and storage. You pay for the cluster tier you choose and the storage you provision, and you scale them independently. Azure DocumentDB doesn't charge request units, per-operation I/O fees, or separate fees for backups, high-performance storage (Premium SSD v2 disks), or text and vector search. As a result, Azure DocumentDB costs at least 40% less than MongoDB Atlas for equivalent compute, and most customers save more than 50% in total when migrating from MongoDB Atlas to Azure DocumentDB (with MongoDB compatibility).

For current prices in your region, see Azure DocumentDB pricing and the Azure pricing calculator.

What you pay for

An Azure DocumentDB bill is made of these items:

Item How it's billed What changes it
Compute Per hour, by cluster tier (vCores and memory), for each shard Scaling up to a larger or scaling down to a smaller tier. Adding shards as the cluster grows increases cost.
Storage Per GiB per month, for the storage you provision on each shard Increasing storage (storage can't be decreased)
High availability Azure DocumentDB uniquely offers the option to selectively enable HA (replica sets), which is recommended only for non production clusters (M10, M20 and M25) and should be enabled for Production tiers (M30 and larger SKUs) Turning high availability on or off
Replicas locally or in other regions Billed as compute and storage in the replica's region Adding or removing cross-region replicas
Data transfer Standard Azure bandwidth rates for networking traffic in and out of the application's virtual network. There are no charge for networking traffic going in and out of the Azure DocumentDB cluster's hosted virtual network to avoid a double charging of networking traffic Location and traffic volume have no impact on networking charges from Azure DocumentDB as they are non existent

What's included at no extra cost

Azure DocumentDB includes these features at no added cost:

  • Backups. Azure DocumentDB automatically backs up your cluster and offers point-in-time restore at no extra charge. The service retains backups for production-grade SKUs (M30 and above) for up to 35 days at no added cost.
  • High-performance storage. The service includes Premium SSD v2 storage for the same price as Premium SSD, at no extra charge. Every Premium SSD v2 disk, no matter how small, is configured with a maximum of 80,000 IOPS and 1.2 GB/s of bandwidth. The attainable throughput of Premium SSD v2 disks is only bound by the maximum throughput and IOPS of the Azure DocumentDB Compute cluster tier. For more information, see High performance SSD v2 disks.
  • Text search and vector search. Integrated text search and vector search, including DiskANN and HNSW vector indexes on paid tiers, don't require separate Compute and Storage for search service or tier.
  • Support. Your existing Azure support plan covers Azure DocumentDB, with no separate support contract.
  • Migration tools. The Azure DocumentDB migration extension and its compatibility assessment are free.
  • Monitoring. Azure DocumentDB platform metrics in Azure Monitor are included. Azure bills diagnostic logs sent to Log Analytics at its standard rates.

By comparison, MongoDB Atlas charges separately for backups, data transfer, and Atlas Search, and sells Atlas support plans separately. See Compare Azure DocumentDB with MongoDB Atlas.

Compute: cluster tiers

The cluster tier sets the vCores and memory for each shard. Every shard in a cluster uses the same tier. You can change tiers at any time without downtime.

Cluster tier vCores Memory (GiB per shard) Typical use
Free Shared Shared Learning, prototypes, and understanding the basic functionalities of Azure DocumentDB (with MongoDB compatibility)
M10 1 2 Development and functional validation
M20 2 4 Development and functional validation for apps that need more than 1 core and 2 GiB of memory
M25 2 (Burstable) 8 Pre-Production and validation
M30 2 8 GiB First production grade SKU
M40 and larger 4 and higher, growing in multiples of 2 16 and higher, growing in multiples of 2 Larger production workloads

Storage

You choose storage for each shard separately from compute, so a workload with a lot of data doesn't need extra vCores. Azure DocumentDB supports up to 32 TB of storage per node. In addition to large disks, Azure DocumentDB also compresses a workload's data on disk.

  • You pay for the storage you provision, not only the storage you use.
  • You can increase storage on an existing cluster, but you can't decrease it.
  • Premium SSD v2 storage is included at no extra charge. The maximum 80,000 IOPS and 1.2 GB/s of bandwidth are available at no added charge and configured by default. So long as the Compute tier is able to push the maximum IOPS and bandwidth, an SSD v2 disk of any storage size get the configured maximum IOPS and bandwidth with no added charge.

High availability

High availability keeps a standby replica of each shard in another availability zone and fails over automatically and instantly.

For development and test, you can turn high availability off and pay only for the primary nodes. MongoDB Atlas clusters are always highly available, so you pay for replicas even in development and test environments without the option to turn off the feature and save on cost.

Shards

Each shard is billed for its compute and storage. A cluster with three shards on the M30 tier, with 512 GiB per shard, pays for three M30 nodes and 1,536 GiB of storage, plus high-availability standbys if you turn high availability on.

Most workloads don't need more than one shard until a database reaches multiple terabytes, because a single node supports up to 32 TB of storage.

Free tier

The Azure DocumentDB free tier gives you a dedicated cluster with 32 GB of storage, free for the lifetime of your Azure account. Each Azure subscription can have one free tier cluster. You can upgrade it to a paid tier in place, keeping your data and connection string. See Azure DocumentDB free tier.

Discounts

  • Reserved capacity. Commit to Azure DocumentDB compute for one or three years to save up to 40% for one year and 60% for a three year commitmemt.
  • Savings plan for databases. Commit to a fixed hourly spend for one year and get lower prices across eligible Azure database services, including Azure DocumentDB.
  • Azure commitments. Azure DocumentDB is a first-party Azure service billed on your Azure invoice.

Estimate your cost

  1. Pick a cluster tier. Start from the vCores and memory your current MongoDB deployment uses. If you're migrating, run a trial migration on representative data to confirm the tier.
  2. Pick storage per shard. Use your current data and index size, plus room to grow.
  3. Decide on shards. Use one shard unless your data is over 32 terabytes in size or your request volume is too large for one shard.
  4. Decide on high availability. Turn it on for production as availability SLAs require HA to be enabled.
  5. Add replicas locally and in other regions if you need them for disaster recovery or local reads.
  6. Apply discounts. Price reserved capacity or the savings plan for steady workloads.
  7. Enter the result in the Azure pricing calculator.

Lower your cost

  • Use the free tier for learning and functional validation.
  • Scale storage, not compute, when only your data is growing.
  • Scale compute down when load drops. You can change cluster tiers without downtime.
  • Buy reserved capacity for production clusters that run all the time.