Compare Azure Event Hubs tiers
This article compares tiers of Azure Event Hubs.
Note
This article compares only features and quotas of tiers. For pricing information, see Azure Event Hubs pricing.
Features
The following table shows the list of features that are available (or not available) in a specific tier of Azure Event Hubs.
| Feature | Basic | Standard | Premium | Dedicated |
|---|---|---|---|---|
| Tenancy | Multitenant | Multitenant | Multitenant with resource isolation | Exclusive single tenant |
| Private link | N/A | Yes | Yes | Yes |
| Customer-managed key (bring your own key) |
N/A | N/A | Yes | Yes |
| Capture | N/A | Priced separately | Included | Included |
| Dynamic partition scale-out | N/A | N/A | Yes | Yes |
| Ingress events | Pay per million events | Pay per million events | Included | Included |
| Runtime audit logs | N/A | N/A | Yes | Yes |
| Availability zone | Yes | Yes | Yes | Yes |
| Geo-disaster | N/A | Yes | Yes | Yes |
| IP firewall | N/A | Yes | Yes | Yes |
Note
Included in the table means the feature is available and there's no separate charge for using it.
Quotas
The following table shows limits that are different for Basic, Standard, Premium, and Dedicated tiers.
Note
- In the table, CU is capacity unit, PU is processing unit, and TU is throughput unit.
- You can configure TUs for a Basic or Standard tier namespace or PUs for a Premium tier namespace.
- When you create a dedicated cluster, one CU is assigned to the cluster. If you enable the Support scaling option while you create the cluster, you can scale out by increasing CUs or scale in by decreasing CUs for the cluster yourself. For step-by-step instructions, see Scale dedicated cluster. For clusters that don't support the Support scaling feature, submit a ticket to adjust CUs for the cluster.
| Limit | Basic | Standard | Premium | Dedicated |
|---|---|---|---|---|
| Maximum size of Event Hubs publication | 256 KB | 1 MB | 1 MB | 1 MB |
| Number of consumer groups per event hub | 1 | 20 | 100 | 1,000 No limit per CU |
| Number of Kafka consumer groups per namespace | NA | 1,000 | 1,000 | 1,000 |
| Number of brokered connections per namespace | 100 | 5,000 | 10,000 per PU For example, if the namespace is assigned 3 PUs, the limit is 30,000. |
100,000 per CU |
| Maximum retention period of event data | 1 day | 7 days | 90 days | 90 days |
| Event storage for retention | 84 GB per TU | 84 GB per TU | 1 TB per PU | 10 TB per CU |
| Maximum TUs or PUs or CUs | 40 TUs | 40 TUs | 16 PUs | 20 CUs |
| Number of partitions per event hub | 32 | 32 | 100 per event hub, but there's a limit of 200 per PU at the namespace level. For example, if a namespace is assigned 2 PUs, the limit for total number of partitions in all event hubs in the namespace is 2 * 200 = 400. |
1,024 per event hub 2,000 per CU |
| Number of namespaces per subscription | 1,000 | 1,000 | 1,000 | 1,000 (50 per CU) |
| Number of event hubs per namespace | 10 | 10 | 100 per PU | 1,000 |
| Capture | N/A | Pay per hour | Included | Included |
| Size of compacted event hub | N/A | 1 GB per partition | 250 GB per partition | 250 GB per partition |
| Size of the schema registry (namespace) in megabytes | N/A | 25 | 100 | 1,024 |
| Number of schema groups in a schema registry or namespace | N/A | 1: excluding the default group | 100 1 MB per schema |
1,000 1 MB per schema |
| Number of schema versions across all schema groups | N/A | 25 | 1,000 | 10,000 |
| Throughput per unit | Ingress: 1 MB/sec or 1000 events per second Egress: 2 MB/sec or 4,096 events per second |
Ingress: 1 MB/sec or 1,000 events per second Egress: 2 MB/sec or 4,096 events per second |
No limits per PU * | No limits per CU * |
* Depends on factors such as resource allocation, number of partitions, and storage.
Note
You can publish events individually or batched. The publication limit (according to SKU) applies regardless of whether it's a single event or a batch. Publishing events larger than the maximum threshold will be rejected.
Related content
For pricing information for each tier, see Azure Event Hubs pricing.