ApsaraDB for MongoDB sharded cluster instances consist of three node types: Mongos, Shard, and ConfigServer. This page lists available specifications for each node type across disk-based and local disk-based instance categories.
The memory shown for each instance type reflects the total allocated memory, including memory used by MongoDB management services, database services, and the operating system, as well as memory reserved for BIOS, kernel, and hypervisor. Available memory is less than the total.
Some instance types were upgraded on May 14, 2024. If the maximum connections for your instance do not match the values below, restart the instance for the upgrade to take effect. For more information, see Notification: Maximum connections upgraded for some ApsaraDB for MongoDB instance types .
A single collection file cannot exceed 16 TB. Exceeding this limit makes the instance unavailable.
Disk-based instances
Disk-based instances support MongoDB 8.0, 7.0, 6.0, 5.0, and 4.4.
For maximum input/output operations per second (IOPS) and throughput values for specific disk types, see Enterprise SSDs (ESSDs) and ESSD AutoPL disks .
Instance type families
Two instance type families are available for disk-based instances:
-
Dedicated disk-based : Higher IOPS ceilings suited for latency-sensitive or I/O-intensive workloads.
-
General-purpose disk-based : Cost-effective for moderate workloads with flexible storage needs.
Mongos node specifications
Mongos nodes route client requests to the appropriate shard. They store no data and have no IOPS, throughput, or storage constraints. The maximum connections listed below are per Mongos node .
Dedicated disk-based
| Specifications | Specification code | Maximum connections (per node) |
|---|---|---|
| 2 vCPU, 8 GB | mdb.shard.4x.large.d | 5,000 |
| 2 vCPU, 16 GB | mdb.shard.8x.large.d | 8,000 |
| 4 vCPU, 8 GB | mdb.shard.2x.xlarge.d | 8,000 |
| 4 vCPU, 16 GB | mdb.shard.4x.xlarge.d | 8,000 |
| 4 vCPU, 32 GB | mdb.shard.8x.xlarge.d | 12,000 |
| 8 vCPU, 16 GB | mdb.shard.2x.2xlarge.d | 12,000 |
| 8 vCPU, 32 GB | mdb.shard.4x.2xlarge.d | 16,000 |
| 8 vCPU, 64 GB | mdb.shard.8x.2xlarge.d | 16,000 |
| 16 vCPU, 32 GB | mdb.shard.2x.4xlarge.d | 16,000 |
| 16 vCPU, 64 GB | mdb.shard.4x.4xlarge.d | 32,000 |
| 16 vCPU, 128 GB | mdb.shard.8x.4xlarge.d | 32,000 |
| 32 vCPU, 64 GB | mdb.shard.2x.8xlarge.d | 32,000 |
| 32 vCPU, 128 GB | mdb.shard.4x.8xlarge.d | 96,000 |
| 32 vCPU, 256 GB | mdb.shard.8x.8xlarge.d | 96,000 |
General-purpose disk-based
| Specifications | Specification code | Maximum connections (per node) |
|---|---|---|
| 2 vCPU, 4 GB | mdb.shard.2x.large.c | 3,000 |
| 4 vCPU, 8 GB | mdb.shard.2x.xlarge.c | 6,000 |
| 8 vCPU, 16 GB | mdb.shard.2x.2xlarge.c | 12,000 |
| 16 vCPU, 32 GB | mdb.shard.2x.4xlarge.c | 16,000 |
| 32 vCPU, 64 GB | mdb.shard.2x.8xlarge.c | 32,000 |
Shard node specifications
Shard nodes store your data. IOPS and throughput are calculated using the formulas below based on the allocated storage size and the instance type's performance tier.
IOPS formula : min{1800 + 50 × Storage space (GB), IOPS cap, Maximum IOPS of the selected disk type}
Throughput formula : min{120 + 0.5 × Storage space (GB), Throughput cap (MB/s), Maximum throughput of the selected disk type}
Dedicated disk-based
| Specifications | Specification code | Maximum connections | IOPS cap | Throughput cap (MB/s) | Storage space |
|---|---|---|---|---|---|
| 2 vCPU, 8 GB | mdb.shard.4x.large.d | 5,000 | 10,000 | 128 | PL1 ESSD: 20–16,000 GB (10 GB step)<br>PL2 ESSD: 470–16,000 GB (10 GB step)<br>PL3 ESSD: 1,270–16,000 GB (10 GB step)<br>ESSD AutoPL disk: 40–16,000 GB (10 GB step) |
| 2 vCPU, 16 GB | mdb.shard.8x.large.d | 8,000 | 10,000 | 128 | Same as above |
| 4 vCPU, 8 GB | mdb.shard.2x.xlarge.d | 8,000 | 20,000 | 192 | Same as above |
| 4 vCPU, 16 GB | mdb.shard.4x.xlarge.d | 8,000 | 20,000 | 192 | Same as above |
| 4 vCPU, 32 GB | mdb.shard.8x.xlarge.d | 12,000 | 20,000 | 192 | Same as above |
| 8 vCPU, 16 GB | mdb.shard.2x.2xlarge.d | 12,000 | 25,000 | 256 | Same as above |
| 8 vCPU, 32 GB | mdb.shard.4x.2xlarge.d | 16,000 | 25,000 | 256 | Same as above |
| 8 vCPU, 64 GB | mdb.shard.8x.2xlarge.d | 16,000 | 25,000 | 256 | Same as above |
| 16 vCPU, 32 GB | mdb.shard.2x.4xlarge.d | 16,000 | 40,000 | 384 | Same as above |
| 16 vCPU, 64 GB | mdb.shard.4x.4xlarge.d | 32,000 | 40,000 | 384 | Same as above |
| 16 vCPU, 128 GB | mdb.shard.8x.4xlarge.d | 32,000 | 40,000 | 384 | Same as above |
| 32 vCPU, 64 GB | mdb.shard.2x.8xlarge.d | 32,000 | 60,000 | 640 | Same as above |
| 32 vCPU, 128 GB | mdb.shard.4x.8xlarge.d | 96,000 | 60,000 | 640 | Same as above |
| 32 vCPU, 256 GB | mdb.shard.8x.8xlarge.d | 96,000 | 60,000 | 640 | Same as above |
| 64 vCPU, 128 GB | mdb.shard.2x.16xlarge.d | 96,000 | 300,000 | 2,048 | Same as above |
| 64 vCPU, 256 GB | mdb.shard.4x.16xlarge.d | 128,000 | 300,000 | 2,048 | Same as above |
| 64 vCPU, 512 GB | mdb.shard.8x.16xlarge.d | 128,000 | 300,000 | 2,048 | Same as above |
The 64 vCPU dedicated disk-based shard node types (128 GB, 256 GB, and 512 GB) are available only in Zones A, B, and C in the Singapore region.
General-purpose disk-based
| Specifications | Specification code | Maximum connections | IOPS cap | Throughput cap (MB/s) |
|---|---|---|---|---|
| 2 vCPU, 4 GB | mdb.shard.2x.large.c | 3,000 | 10,500 | 128 |
| 2 vCPU, 8 GB | mdb.shard.4x.large.c | 3,000 | 10,500 | 128 |
| 4 vCPU, 8 GB | mdb.shard.2x.xlarge.c | 6,000 | 21,000 | 192 |
| 4 vCPU, 16 GB | mdb.shard.4x.xlarge.c | 6,000 | 21,000 | 192 |
| 8 vCPU, 16 GB | mdb.shard.2x.2xlarge.c | 12,000 | 26,250 | 256 |
| 8 vCPU, 32 GB | mdb.shard.4x.2xlarge.c | 16,000 | 26,250 | 256 |
| 16 vCPU, 32 GB | mdb.shard.2x.4xlarge.c | 16,000 | 42,000 | 384 |
| 16 vCPU, 64 GB | mdb.shard.4x.4xlarge.c | 32,000 | 42,000 | 384 |
| 32 vCPU, 64 GB | mdb.shard.2x.8xlarge.c | 32,000 | 50,000 | 640 |
General-purpose disk-based instances are available in the following regions:
-
China (Hangzhou): Zone J
-
China (Shanghai): Zones B, L, and N
-
China (Beijing): Zones H and K
-
China (Shenzhen): Zone D
-
China (Hong Kong): Zones B and C
-
Singapore: Zone A
-
US (Silicon Valley): Zones A and B
-
US (Virginia): Zones A and B
For multi-zone deployment, confirm that the destination region supports the selected instance type.
Disk encryption is not supported for General-purpose disk-based instance types.
General-purpose disk-based instances are not available in all regions. If the required instance type is not listed in the console for your region, submit a ticket to request access.
ConfigServer node specifications
ConfigServer nodes store cluster metadata and shard key ranges.
Dedicated disk-based
| Specifications | Specification code | Maximum connections | IOPS cap | Throughput cap (MB/s) | Storage space |
|---|---|---|---|---|---|
| 4 vCPU, 8 GB | mdb.shard.2x.xlarge.d | 8,000 | 20,000 | 192 | PL1 ESSD: 20–16,000 GB (10 GB step)<br>PL2 ESSD: 470–16,000 GB (10 GB step)<br>PL3 ESSD: 1,270–16,000 GB (10 GB step)<br>ESSD AutoPL disk: 40–16,000 GB (10 GB step) |
General-purpose disk-based
| Specifications | Specification code | Maximum connections | IOPS cap | Throughput cap (MB/s) |
|---|---|---|---|---|
| 2 vCPU, 4 GB | mdb.shard.2x.large.c | 3,000 | 10,500 | 128 |
| 4 vCPU, 8 GB | mdb.shard.2x.xlarge.c | 6,000 | 21,000 | 192 |
Local disk-based instances
Local disk-based instances use physical disks attached to the host and support MongoDB 4.2, 4.0, and 3.4 (discontinued).
Mongos node specifications
General-purpose local disk-based
| Specifications | Specification code | Maximum connections (per node) |
|---|---|---|
| 1 vCPU, 2 GB | dds.mongos.mid | 1,000 |
| 2 vCPU, 4 GB | dds.mongos.standard | 3,000 |
| 4 vCPU, 8 GB | dds.mongos.large | 6,000 |
| 8 vCPU, 16 GB | dds.mongos.xlarge | 9,000 |
| 8 vCPU, 32 GB | dds.mongos.2xlarge | 16,000 |
| 16 vCPU, 64 GB | dds.mongos.4xlarge | 32,000 |
Shard node specifications
General-purpose local disk-based
| Specifications | Specification code | Maximum connections | Maximum IOPS | Storage space |
|---|---|---|---|---|
| 1 vCPU, 2 GB | dds.shard.mid | 8,000 | 8,000 | 10–2,000 GB (10 GB step) |
| 2 vCPU, 4 GB | dds.shard.standard | 8,000 | 8,000 | 10–2,000 GB (10 GB step) |
| 4 vCPU, 8 GB | dds.shard.large | 8,000 | 8,000 | 10–2,000 GB (10 GB step) |
| 8 vCPU, 16 GB | dds.shard.xlarge | 8,000 | 8,000 | 10–2,000 GB (10 GB step) |
| 8 vCPU, 32 GB | dds.shard.2xlarge | 16,000 | 14,000 | 10–2,000 GB (10 GB step) |
| 16 vCPU, 64 GB | dds.shard.4xlarge | 32,000 | 16,000 | 10–2,000 GB (10 GB step) |
Dedicated local disk-based
| Specifications | Specification code | Maximum connections | Maximum IOPS | Storage space |
|---|---|---|---|---|
| 2 vCPU, 16 GB | dds.shard.sn8.xlarge.3 | 8,000 | 8,000 | 10–250 GB (10 GB step) |
| 4 vCPU, 32 GB | dds.shard.sn8.2xlarge.3 | 8,000 | 9,000 | 10–500 GB (10 GB step) |
| 8 vCPU, 64 GB | dds.shard.sn8.4xlarge.3 | 16,000 | 18,000 | 10–1,000 GB (10 GB step) |
| 16 vCPU, 128 GB | dds.shard.sn8.8xlarge.3 | 32,000 | 36,000 | 10–2,000 GB (10 GB step) |
| 32 vCPU, 256 GB | dds.shard.sn8.16xlarge.3 | 64,000 | 72,000 | 10–3,000 GB (10 GB step) |
ConfigServer node specifications
General-purpose local disk-based
| Specifications | Specification code | Maximum connections | Maximum IOPS | Storage space |
|---|---|---|---|---|
| 1 vCPU, 2 GB | dds.cs.mid | 8,000 | 8,000 | 20–1,270 GB (10 GB step) |
The storage space for a shard node is allocated to each individual node. For example, if the storage for a shard component is 20 GB, each primary node and secondary node in that shard has 20 GB of storage.
To choose storage based on IOPS requirements, see Select storage space based on IOPS .