Merge pull request #224321 from jonels-msft/cdbpg-burstable

Burstable compute

Author: v-dirichards

articles/cosmos-db/postgresql/TOC.yml

@@ -103,6 +103,8 @@
       items:
         - name: Scheduled maintenance
           href: concepts-maintenance.md
+        - name: Burstable compute
+          href: concepts-burstable-compute.md
         - name: Upgrade
           href: concepts-upgrade.md
     - name: Business continuity

articles/cosmos-db/postgresql/concepts-burstable-compute.md

@@ -0,0 +1,46 @@
+---
+title: Burstable compute - Azure Cosmos DB for PostgreSQL
+description: Definition and workloads of burstable compute.
+ms.author: jonels
+author: jonels-msft
+ms.service: cosmos-db
+ms.subservice: postgresql
+ms.topic: conceptual
+ms.date: 01/18/2023
+---
+
+# Burstable compute
+
+[!INCLUDE [PostgreSQL](../includes/appliesto-postgresql.md)]
+
+Burstable compute on a single node cluster is ideal for workloads like
+development environments or small databases that don't need the full
+performance of the node's CPU continuously. These workloads typically have
+burstable performance requirements.
+
+The Azure burstable compute options allow you to configure a single node with
+baseline performance that can build up credits when it's using less than its
+baseline. When the node has accumulated credits, the node can burst above the
+baseline when your workload requires higher CPU performance. You can use the
+**CPU credits remaining** and **CPU credits consumed** metrics to track
+accumulated and used credits respectively.
+
+> [!IMPORTANT]
+>
+> Scaling for burstable compute is available, but only in certain combinations:
+>
+> * You can change 1 vCore burstable to 2 vCore burstable, and the other way
+>   around.
+> * You can convert a burstable (1 or 2 vCores) single node configuration to
+>   either
+>   1. a single node cluster with regular (non-burstable) compute; or
+>   2. a multi-node cluster.
+>
+> You also can't go back from a regular (non-burstable) single node to
+> burstable compute.
+
+**Next steps**
+
+* See the [limits and limitations](reference-limits.md#burstable-compute) of
+  burstable compute.
+* Review available [cluster metrics](concepts-monitoring.md#metrics).

articles/cosmos-db/postgresql/concepts-monitoring.md

@@ -7,7 +7,7 @@ ms.service: cosmos-db
 ms.subservice: postgresql
 ms.custom: ignite-2022
 ms.topic: conceptual
-ms.date: 09/27/2022
+ms.date: 01/18/2023
 ---
 
 # Monitor and tune Azure Cosmos DB for PostgreSQL
@@ -56,6 +56,8 @@ These metrics are available for nodes:
 |---|---|---|---|
 |active_connections|Active Connections|Count|The number of active connections to the server.|
 |apps_reserved_memory_percent|Reserved Memory Percent|Percent|Calculated from the ratio of Committed_AS/CommitLimit as shown in /proc/meminfo.|
+|cpu_credits_consumed|CPU credits consumed|Credits|Total number of credits consumed by the node. (Only available when burstable compute is provisioned on the node.)|
+|cpu_credits_remaining|CPU credits remaining|Credits|Total number of credits available to burst. (Only available when burstable compute is provisioned on the node.)|
 |cpu_percent|CPU percent|Percent|The percentage of CPU in use.|
 |iops|IOPS|Count|See the [IOPS definition](../../virtual-machines/premium-storage-performance.md#iops) and [Azure Cosmos DB for PostgreSQL throughput](resources-compute.md)|
 |memory_percent|Memory percent|Percent|The percentage of memory in use.|

articles/cosmos-db/postgresql/product-updates.md

@@ -7,7 +7,7 @@ ms.custom: mvc
 ms.service: cosmos-db
 ms.subservice: postgresql
 ms.topic: conceptual
-ms.date: 01/11/2023
+ms.date: 01/23/2023
 ---
 
 # Product updates for Azure Cosmos DB for PostgreSQL
@@ -22,6 +22,11 @@ Updates that don’t directly affect the internals of a cluster are rolled out g
 
 Updates that change cluster internals, such as installing a [new minor PostgreSQL version](https://www.postgresql.org/developer/roadmap/), are delivered to existing clusters as part of the next [scheduled maintenance](concepts-maintenance.md) event. Such updates are available immediately to newly created clusters.
 
+### January 2023
+
+* General availability: 1- and 2-vCore [burstable compute](concepts-burstable-compute.md) options for single-node clusters (for dev/test and smaller workloads).
+	* Available in all supported regions. See compute and storage details [here](resources-compute.md#single-node-cluster).
+
 ### December 2022
 
 * General availability: Azure Cosmos DB for PostgreSQL is now available in the Sweden Central and Switzerland West regions.

articles/cosmos-db/postgresql/reference-limits.md

@@ -35,6 +35,9 @@ to keep nodes healthy:
    * 300 for 0-3 vCores
    * 500 for 4-15 vCores
    * 1000 for 16+ vCores
+* Maximum connections per node with burstable compute
+   * 20 for 1 vCore burstable
+   * 40 for 2 vCores burstable
 
 The connection limits above are for *user* connections (`max_connections` minus
 `superuser_reserved_connections`). We reserve extra connections for
@@ -74,6 +77,20 @@ Azure Cosmos DB for PostgreSQL. If you do need more vCores for a region in your
 subscription, see how to [adjust compute
 quotas](howto-compute-quota.md).
 
+### Burstable compute
+
+In Azure Cosmos DB for PostgreSQL clusters with [burstable
+compute](concepts-burstable-compute.md) enabled, the following features are
+currently **not supported**:
+
+* Accelerated networking
+* Local caching
+* PostgreSQL and Citus version upgrades
+* PostgreSQL 11 support
+* Read replicas
+* High availability
+* The [azure_storage](howto-ingest-azure-blob-storage.md) extension
+
 ## PostgreSQL
 
 ### Database creation

articles/cosmos-db/postgresql/resources-compute.md

@@ -7,29 +7,30 @@ ms.service: cosmos-db
 ms.subservice: postgresql
 ms.custom: ignite-2022
 ms.topic: conceptual
-ms.date: 07/08/2022
+ms.date: 01/27/2023
 ---
 
 # Azure Cosmos DB for PostgreSQL compute and storage
 
 [!INCLUDE [PostgreSQL](../includes/appliesto-postgresql.md)]
+
+Compute resources are provided as vCores, which represent the logical CPU of
+the underlying hardware. The storage size for provisioning refers to the
+capacity available to the coordinator and worker nodes in your cluster. The
+storage includes database files, temporary files, transaction logs, and the
+Postgres server logs.
+
+## Multi-node cluster
 
-You can select the compute and storage settings independently for
-worker nodes and the coordinator node in a cluster.
-Compute resources are provided as vCores, which represent
-the logical CPU of the underlying hardware. The storage size for
-provisioning refers to the capacity available to the coordinator
-and worker nodes in your cluster. The storage
-includes  database files, temporary files, transaction logs, and
-the Postgres server logs.
+You can select the compute and storage settings independently for worker nodes
+and the coordinator node in a multi-node cluster.
 
 | Resource              | Worker node                    | Coordinator node         |
 |-----------------------|--------------------------------|--------------------------|
 | Compute, vCores       | 4, 8, 16, 32, 64, 96, 104      | 4, 8, 16, 32, 64, 96     |
 | Memory per vCore, GiB | 8                              | 4                        |
 | Storage size, TiB     | 0.5, 1, 2                      | 0.5, 1, 2                |
 | Storage type          | General purpose (SSD)          | General purpose (SSD)    |
-| IOPS                  | Up to 3 IOPS/GiB               | Up to 3 IOPS/GiB         |
 
 The total amount of RAM in a single node is based on the
 selected number of vCores.
@@ -49,36 +50,62 @@ available to each worker and coordinator node.
 
 | Storage size, TiB | Maximum IOPS |
 |-------------------|--------------|
-| 0.5               | 1,536        |
-| 1                 | 3,072        |
-| 2                 | 6,148        |
+| 0.5               | 2,300        |
+| 1                 | 5,000        |
+| 2                 | 7,500        |
 
 For the entire cluster, the aggregated IOPS work out to the
 following values:
 
 | Worker nodes | 0.5 TiB, total IOPS | 1 TiB, total IOPS | 2 TiB, total IOPS |
 |--------------|---------------------|-------------------|-------------------|
-| 2            | 3,072               | 6,144             | 12,296            |
-| 3            | 4,608               | 9,216             | 18,444            |
-| 4            | 6,144               | 12,288            | 24,592            |
-| 5            | 7,680               | 15,360            | 30,740            |
-| 6            | 9,216               | 18,432            | 36,888            |
-| 7            | 10,752              | 21,504            | 43,036            |
-| 8            | 12,288              | 24,576            | 49,184            |
-| 9            | 13,824              | 27,648            | 55,332            |
-| 10           | 15,360              | 30,720            | 61,480            |
-| 11           | 16,896              | 33,792            | 67,628            |
-| 12           | 18,432              | 36,864            | 73,776            |
-| 13           | 19,968              | 39,936            | 79,924            |
-| 14           | 21,504              | 43,008            | 86,072            |
-| 15           | 23,040              | 46,080            | 92,220            |
-| 16           | 24,576              | 49,152            | 98,368            |
-| 17           | 26,112              | 52,224            | 104,516           |
-| 18           | 27,648              | 55,296            | 110,664           |
-| 19           | 29,184              | 58,368            | 116,812           |
-| 20           | 30,720              | 61,440            | 122,960           |
-
-**Next steps**
+| 2            | 4,600               | 10,000            | 15,000            |
+| 3            | 6,900               | 15,000            | 22,500            |
+| 4            | 9,200               | 20,000            | 30,000            |
+| 5            | 11,500              | 25,000            | 37,500            |
+| 6            | 13,800              | 30,000            | 45,000            |
+| 7            | 16,100              | 35,000            | 52,500            |
+| 8            | 18,400              | 40,000            | 60,000            |
+| 9            | 20,700              | 45,000            | 67,500            |
+| 10           | 23,000              | 50,000            | 75,000            |
+| 11           | 25,300              | 55,000            | 82,500            |
+| 12           | 27,600              | 60,000            | 90,000            |
+| 13           | 29,900              | 65,000            | 97,500            |
+| 14           | 32,200              | 70,000            | 105,000           |
+| 15           | 34,500              | 75,000            | 112,500           |
+| 16           | 36,800              | 80,000            | 120,000           |
+| 17           | 39,100              | 85,000            | 127,500           |
+| 18           | 41,400              | 90,000            | 135,000           |
+| 19           | 43,700              | 95,000            | 142,500           |
+| 20           | 46,000              | 100,000           | 150,000           |
+
+## Single node cluster
+
+Single-node cluster resource options differ between [burstable
+compute](concepts-burstable-compute.md) and regular compute.
+
+**Burstable compute**
+
+| Resource | Resource value |
+|----------|----------------|
+| Burstable compute, vCores | 1, 2 |
+| Burstable compute memory per vCore, GiB | 4 |
+| Storage size, GiB | 32, 64, 128 |
+| Storage IOPS | Up to 500 |
+| Storage type | General purpose (SSD) |
+
+**Regular compute**
+
+| Resource | Resource value |
+|----------|----------------|
+| Compute, vCores | 2, 4, 8, 16, 32, 64 |
+| Compute memory per vCore, GiB | 4 |
+| Storage size, GiB (IOPS, up to) | 128 (500), 256 (1,100), 512 (2,300), 1024† (5,000), 2048† (7,500) |
+| Storage type | General purpose (SSD) |
+
+† 1024 GiB and 2048 GiB are supported with 8 vCores or greater.
+
+## Next steps
 
 * Learn how to [create a cluster in the portal](quickstart-create-portal.md)
 * Change [compute quotas](howto-compute-quota.md) for a subscription and region