You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This article gives a high-level overview of the Azure Local Rack Aware Clustering feature. Azure Local Rack Aware Cluster is an advanced architecture designed to enhance fault tolerance and data distribution within an Azure Local instance. This architecture enables you to cluster nodes that are strategically placed across two physical racks in two different rooms or buildings with high bandwidth and low latency. An example of Rack Aware Cluster could be the use in a manufacturing plant to minimize the risk of data loss or downtime in the event of a rack-level failure.
15
+
This article gives a high-level overview of the Azure Local Rack Aware Clustering feature including its benefits and use cases. The article also details the supported configurations and deployment requirements for rack aware clustering.
16
16
17
-
The diagram shows a cluster of two racks of servers with top-of-rack (Tor) switches connected between rooms. Each rack functions as a local availability zone across layers from the operating system to Azure Local management, including Azure Local VMs enabled by Azure Arc and Azure Kubernetes Service (AKS).
:::image type="content" source="media/rack-aware-clustering-overview/rack-aware-cluster-architecture.png" alt-text="Diagram of rack aware cluster architecture." lightbox="media/rack-aware-clustering-overview/rack-aware-cluster-architecture.png":::
19
+
## About Rack Aware Cluster
20
20
21
-
This direct connection between racks supports a single storage pool, with Rack Aware Clusters guaranteeing data copy distribution evenly between the two racks. This design is valuable in environments that require high availability and disaster recovery capabilities. Even if an entire rack encounters an issue, the other rack maintains the integrity and accessibility of the data.
21
+
Azure Local Rack Aware Cluster is an advanced architecture designed to enhance fault tolerance and data distribution within an Azure Local instance. This architecture enables you to cluster nodes that are strategically placed across two physical racks in two different rooms or buildings with high bandwidth and low latency.
22
22
23
-
To support the synchronous replications between racks, a dedicated storage network intent is required to secure the bandwidth and low latency for storage traffic. The required round-trip latency between the two racks should be 1 ms or less.
23
+
The following diagram shows a cluster of two racks of servers with top-of-rack (TOR) switches connected between rooms. Each rack functions as a local availability zone across layers from the operating system to Azure Local management, including Azure Local VMs enabled by Azure Arc and Azure Kubernetes Service (AKS).
24
24
25
-
For detailed networking requirements, see [Rack Aware Clustering network design](rack-aware-clustering-network-design.md).
25
+
:::image type="content" source="media/rack-aware-clustering-overview/rack-aware-cluster-architecture.png" alt-text="Diagram of rack aware cluster architecture." lightbox="media/rack-aware-clustering-overview/rack-aware-cluster-architecture.png":::
26
26
27
-
## Requirements and supported configurations
27
+
This direct connection between racks supports a single storage pool, with Rack Aware Clusters guaranteeing data copy distribution evenly between the two racks.
28
28
29
-
All [system requirements for Azure Local](../concepts/system-requirements-23h2.md) apply to rack aware clusters. Other requirements for rack-aware clusters include:
29
+
This design is valuable in environments that require high availability and disaster recovery capabilities. Even if an entire rack encounters an issue, the other rack maintains the integrity and accessibility of the data.
30
30
31
-
- Data drives must be all-flash, either Nonvolatile Memory Express (NVMe) or solid-state drives (SSD).
31
+
To support the synchronous replications between racks, a dedicated storage network intent is required to secure the bandwidth and low latency for storage traffic. The required round-trip latency between the two racks should be 1 ms or less.
32
32
33
-
- Azure Local Rack Aware Cluster supports only two local availability zones, with a maximum of four machines in each zone. The two zones must contain an equal number of machines. See the table for supported configurations with volume resiliency settings.
33
+
For detailed networking requirements, see [Rack Aware Clustering network design](./rack-aware-clustering-network-design-switch-configuration.md).
34
34
35
-
| Machines in two zones | Volume resiliency | Infrastructure volumes | Workload volumes |
- New greenfield deployments are supported only - no conversion from standard clusters.
37
+
The main benefits of Azure Local Rack Aware Cluster include:
43
38
44
-
-To facilitate synchronous replications between racks, a dedicated storage network is essential to ensure adequate bandwidth and low latency for storage traffic. The round-trip latency requirement between two racks should be 1 millisecond or less. The necessary bandwidth can be calculated based on the cluster size and the network interface card (NIC) speed as follows:
39
+
-**High availability**: By distributing data across two or more racks, the system can withstand the failure of an entire rack without losing access to data. This is particularly important for mission-critical applications that require continuous uptime.
45
40
46
-
| Machines in zone | NIC speed | Storage ports | Bandwidth required |
47
-
| -- | -- | -- | -- |
48
-
| 1 | 10 | 2 | 20 GbE |
49
-
| 2 | 10 | 2 | 40 GbE |
50
-
| 3 | 10 | 2 | 60 GbE |
51
-
| 4 | 10 | 2 | 80 GbE |
52
-
| 1 | 25 | 2 | 50 GbE |
53
-
| 2 | 25 | 2 | 100 GbE |
54
-
| 3 | 25 | 2 | 150 GbE |
55
-
| 4 | 25 | 2 | 200 GbE |
41
+
-**Improved performance**: The architecture allows for better load balancing and resource utilization, as workloads can be distributed across multiple racks.
56
42
43
+
-**Simplified management**: Azure Local Rack Aware Clustering provides a unified management interface for both racks, making it easier to monitor and maintain the system.
57
44
58
-
> [!NOTE]
59
-
> Adding a machine to a cluster (`add-node` command) isn't supported in Azure Local version 2504.
45
+
-**Cost-effective**: By leveraging existing infrastructure and providing a single storage pool, organizations can reduce costs associated with hardware and maintenance.
60
46
61
-
## Storage design
62
47
63
-
Storage Spaces Direct is used to create a single storage pool that aggregates the disk capacity from all machines. For a 1+1 configuration, two volumes are created—one on each machine—with a two-way mirror that respects the rack fault domain, ensuring two copies of data are available in the cluster, one in each rack. In a 2+2 configuration, four volumes are created—one on each machine—with a two-way mirror that also respects the rack fault domain, providing one copy of data in each rack. With a two-way mirror, the total usable capacity of the storage pool is 50%. With a two-way mirror, the system can handle one type of failure at a time. This means each cluster can support the failure of either one rack, one machine, or one disk without losing data.
48
+
## Use cases
64
49
65
-
Only two-way mirror volumes are supported. Three-way mirror volumes aren't supported.
50
+
The use cases for Azure Local Rack Aware Cluster include:
66
51
67
-
## Azure Local virtual machines
52
+
- Use rack aware cluster in a high security environment for mission-critical applications.
53
+
- Use rack aware cluster in a manufacturing plant to minimize the risk of data loss or downtime in the event of a rack-level failure.
54
+
- Barbara to provide more information on use cases.
68
55
69
-
This section deals with Azure Local virtual machines (VMs).
70
56
71
-
### VM prerequisites
57
+
##Requirements and supported configurations
72
58
73
-
Before you create an Azure Local VM for Rack Aware Clustering, make sure that all prerequisites listed in [Create Azure Local virtual machines enabled by Azure Arc](../manage/create-arc-virtual-machines.md)are met.
59
+
All [system requirements for Azure Local](../concepts/system-requirements-23h2.md)apply to rack aware clusters.
74
60
75
-
### Create a VM in a specific availability zone
61
+
Other requirements for rack aware clusters include:
76
62
77
-
Once availability zones are set up in Rack Aware Cluster, Azure Local VMs can be created on a specific availability zone to reduce latency, improve performance, ensure redundancy, and meet compliance requirements.
63
+
-**Drive requirements**: Data drives must be all-flash, either non volatile memory express (NVMe) or solid-state drives (SSD).
78
64
79
-
When you create a VM in a specific availability zone, the default option is without strict placement. The VM is created on a machine within the specified zone (`--zone`). If all machines within a zone are down, the VM can migrate to another machine outside of the zone. However, the VM attempts to failback to a machine within the original zone when that availability zone recovers.
65
+
-**Availability zone requirements**: Rack aware cluster supports only two local availability zones, with a maximum of four machines in each zone. The two zones must contain an equal number of machines.
80
66
81
-
> [!NOTE]
82
-
> If no zone is specified when creating an Azure Local VM, the system will automatically choose an optimal placement within the Rack Aware Cluster.
67
+
-**Supported configurations**:
83
68
84
-
1. Connect to a machine on your Azure Local instance.
85
-
1. Sign in and set the subscription:
69
+
- The following table summarizes the supported configurations with volume resiliency settings.
86
70
87
-
:::image type="content" source="media/rack-aware-clustering-overview/rack-aware-cluster-vm-sign-in.png" alt-text="Screenshot showing VM sign in." lightbox="media/rack-aware-clustering-overview/rack-aware-cluster-vm-sign-in.png":::
71
+
| Machines in two zones | Volume resiliency | Infra volumes | Workload volumes |
-**Replication requirements**: To facilitate synchronous replications between racks, a dedicated storage network is essential to ensure adequate bandwidth and low latency for storage traffic. The round-trip latency requirement between two racks should be 1 millisecond or less. The necessary bandwidth can be calculated based on the cluster size and the network interface card (NIC) speed as follows:
107
81
108
-
1. Run the following command:
82
+
| Machines in zone | NIC speed | Storage ports | Bandwidth required |
1. Verify that the VM was created in the desired zone. Look under `placementProfile` in the output.
93
+
-**Adding a machine to a cluster** using the `add-node` command isn't supported in this release.
115
94
116
-
```azurecli
117
-
"placementProfile": {
118
-
"strictPlacementPolicy": null,
119
-
"zone": "local-zone"
120
-
},
121
-
```
95
+
## Storage design
122
96
123
-
### Create a VM in a specific availability zone (strict placement enabled)
97
+
Storage Spaces Direct is used to create a single storage pool that aggregates the disk capacity from all machines.
124
98
125
-
To create a VM in a specific availability zone and ensure it stays within that availability zone, you can specify another flag `--strict-placement` to `true`. The VM is created on a node within the specified zone (`--zone`). If all machines within a zone are down, the VM also goes down. Only enable strict placement if the VM isn't required to be always running or needs to adhere to a specific availability zone due to compliance requirements.
99
+
- Only two-way mirror volumes are supported. Three-way mirror volumes aren't supported.
126
100
127
-
1. Run the following command:
101
+
- For a *1+1* configuration, two volumes are created — one on each machine — with a two-way mirror that respects the rack fault domain, ensuring two copies of data are available in the cluster, one in each rack.
102
+
- In a *2+2* configuration, four volumes are created — one on each machine — with a two-way mirror that also respects the rack fault domain, providing one copy of data in each rack.
1. Verify that the VM has strict placement enabled and was created in the desired zone. Look under `placementProfile` in the output.
106
+
- The total usable capacity of the storage pool is 50%.
107
+
- The system can handle one type of failure at a time. This means each cluster can support the failure of either one rack, one machine, or one disk without losing data.
134
108
135
-
```azurecli
136
-
"placementProfile": {
137
-
"strictPlacementPolicy": true,
138
-
"zone": "local-zone"
139
-
},
140
-
```
141
109
142
110
## Next steps
143
111
144
-
- See [Rack Aware Cluster networking](rack-aware-clustering-network-design.md).
112
+
- See [Rack Aware Cluster networking](./rack-aware-clustering-network-design-switch-configuration.md).
This article gives a high-level overview of the Azure Local Rack Aware Clustering feature including its benefits and use cases. The article also details the supported configurations and deployment requirements for rack aware clustering.
15
+
This article describes how to place an Azure Local virtual machine (VM) on a Rack Aware Cluster.
## About Rack Aware Cluster and virtual machines on Azure Local
19
+
## About Rack Aware Cluster and Azure Local VMs
20
20
21
21
Azure Local Rack Aware Cluster is an advanced architecture designed to enhance fault tolerance and data distribution within an Azure Local instance. This architecture enables you to cluster nodes that are strategically placed across two physical racks in two different rooms or buildings with high bandwidth and low latency. An example of Rack Aware Cluster could be the use in a manufacturing plant to minimize the risk of data loss or downtime in the event of a rack-level failure.
22
22
23
+
## Prerequisites
23
24
24
-
### VM prerequisites
25
+
Before you create an Azure Local VM for a rack aware cluster, make sure that all prerequisites listed in [Create Azure Local virtual machines enabled by Azure Arc](../manage/create-arc-virtual-machines.md) are met.
25
26
26
-
Before you create an Azure Local VM for Rack Aware Clustering, make sure that all prerequisites listed in [Create Azure Local virtual machines enabled by Azure Arc](../manage/create-arc-virtual-machines.md) are met.
27
+
## Create a VM in a specific availability zone without strict placement
27
28
28
-
### Create a VM in a specific availability zone
29
+
Once you have set up availability zones in a rack aware cluster, you can create Azure Local VMs on a specific availability zone to reduce latency, improve performance, ensure redundancy, and meet compliance requirements.
29
30
30
-
Once availability zones are set up in Rack Aware Cluster, Azure Local VMs can be created on a specific availability zone to reduce latency, improve performance, ensure redundancy, and meet compliance requirements.
31
+
When you create a VM in a specific availability zone, the default option is without strict placement. The VM is created on a machine within the specified zone (`--zone`).
31
32
32
-
When you create a VM in a specific availability zone, the default option is without strict placement. The VM is created on a machine within the specified zone (`--zone`). If all machines within a zone are down, the VM can migrate to another machine outside of the zone. However, the VM attempts to failback to a machine within the original zone when that availability zone recovers.
33
+
If all machines within a zone are down, the VM can migrate to another machine outside of the zone. However, the VM attempts to fail back to a machine within the original zone when that availability zone recovers.
33
34
34
35
> [!NOTE]
35
-
> If no zone is specified when creating an Azure Local VM, the system will automatically choose an optimal placement within the Rack Aware Cluster.
36
+
> If no zone is specified when creating an Azure Local VM, the system automatically chooses an optimal placement within the rack aware cluster.
36
37
37
38
1. Connect to a machine on your Azure Local instance.
38
39
1. Sign in and set the subscription:
@@ -73,7 +74,7 @@ When you create a VM in a specific availability zone, the default option is with
73
74
},
74
75
```
75
76
76
-
### Create a VM in a specific availability zone (strict placement enabled)
77
+
## Create a VM in a specific availability zone with strict placement enabled
77
78
78
79
To create a VM in a specific availability zone and ensure it stays within that availability zone, you can specify another flag `--strict-placement` to `true`. The VM is created on a node within the specified zone (`--zone`). If all machines within a zone are down, the VM also goes down. Only enable strict placement if the VM isn't required to be always running or needs to adhere to a specific availability zone due to compliance requirements.
0 commit comments