Creating a stretched cluster. Dell AX-750, AX-640, AX-6515, AX-740XD, AX-7525, AX-650


Add to my manuals
76 Pages

advertisement

Creating a stretched cluster. Dell AX-750, AX-640, AX-6515, AX-740XD, AX-7525, AX-650 | Manualzz

4

Creating a stretched cluster

This chapter presents the following topics:

Topics:

Introduction

Test-Cluster

Cluster creation

Volumes

Storage efficiency

Test-SRTopology

Introduction

This section outlines the steps that are needed for configuring a stretched cluster. Complete the network configuration on all nodes for the network topology applicable to you. A sample IP address schema is provided for both supported network topologies in the previous section of this guide. Consider these points before you begin:

● Ensure management IPs of all nodes are reachable from any host

● Ensure static routes are configured on all hosts for inter-site communication using the Replica network

● Ensure all nodes from Site A can reach corresponding Replica IPs on Site B using the Replica path

Test-Cluster

Test-Cluster is a fundamental function that is needed to ensure that the cluster to be created meets Microsoft recommendations regarding Failover Clustering. It also ensures that your hardware and settings are compatible. Run Test-

Cluster with all nodes and include All Tests (namely, 'Storage Spaces Direct', 'Inventory', 'Network' and 'System Configuration').

Ensure there are test-cluster passes without warnings for the 'High Throughput Configuration', while 'Basic Configuration' will receive a warning as mentioned in the previous section of this guide.

Cluster creation

This section looks at creating a cluster using PowerShell cmdlets.

Manual cluster creation

Once Test-Cluster completes successfully, use the New-Cluster cmdlet to create a new stretched cluster. Because the nodes specified are part of different IP schemas, Enable-ClusterS2D understands that the cluster is part of a multi-site topology. It automatically creates two storage pools and corresponding ClusterPerformanceHistory volumes and their replica volumes.

After a cluster is created, you will see a warning similar to the following example. This is an expected behavior.

No matching network interface found for resource 'Cluster IP Address 172.18.160.160' IP address '192.168.200.100' (return code was '5035'). If your cluster nodes span different subnets, this may be normal.

Configure cluster witness and Enable Storage Spaces Direct on the cluster.

NOTE: Cluster witness can be either on a tertiary site or on Azure Cloud. Ensure that the "Storage Replica" feature is installed on all nodes in the cluster.

16 Creating a stretched cluster

If Sites and Services with IP Subnets are configured on Active Directory, Failover Cluster Manager correctly shows a node to

Site mapping, under Cluster Name >> Nodes .

The following is a sample image of IP subnets defined in an Active Directory:

Figure 5. IP subnets in an Active Directory

If both sites are in the same IP network, use the New-ClusterFaultDomain cmdlet to define the two site names. Site names defined using New-ClusterFaultDomain override the names given in Active Directory.

The following is a sample image of how sites appears in Failover Cluster Manager:

Figure 6. Sites in Failover Cluster Manager

Once a cluster is created, use Failover Cluster Manager to rename the cluster networks.

Creating a stretched cluster 17

Figure 7. Cluster networks

Volumes

Replication-enabled volumes can be created using a combination of PowerShell and Failover Cluster Manager or by using

Windows Admin Center.

NOTE: Install Storage Replica Module for Windows PowerShell (RSAT-Storage-Replica) on the management node with

Desktop Experience that is used for installing Windows Admin Center and Failover Cluster Manager to access the cluster.

NOTE: Site-to-site volume encryption uses additional CPU resources to encrypt data, and could potentially cause cluster performance-related issues. Microsoft and Dell recommend disabling site-to-site volume encryption unless required.

For each replica-enabled volume, you need a corresponding log volume on both sites (with a minimum of 8 GB in size) and an equivalent replica volume on the secondary site. The log volume is used to serialize writes for replication.

The following table shows the volumes that are must create a 1 TB replica volume:

Table 6. Volumes in a 1 TB replica volume

Site A

VolumeA

VolumeA-Log

Size

1 TB

40 GB

VolumeB

VolumeB-Log

500 GB

40 GB

Site B

VolumeA-Replica

VolumeA-Replica-Log

VolumeB-Replica

VolumeB-Replica-Log

Size

1 TB

40 GB

500 GB

40 GB

Once a replicated volume is created, go to Storage Replica > Partnership > Settings > Modify Partnership settings to verify the status of the replica traffic encryption.

18 Creating a stretched cluster

See

Appendix A

for the correct PowerShell Cmdlets and Failover Cluster Manager steps to create the volumes shown in this table. It is recommended that you create two-way mirrors for all volumes to improve write performance and capacity efficiency.

NOTE: For Asynchronous Replication, the RPO can be set as low as 30 s.

For a planned site failure, when the volume replication direction is reversed, the disk reservations on the secondary site for

Replica Volume and Replica-log volumes are removed and moved to the primary site. Source Data and Source Log volumes are given the disk reservations and become active on the secondary site. After 10 minutes, the virtual machines residing on the primary site associated with the migrated volume automatically Live Migrate to the secondary site.

Storage efficiency

Due to high I/Os on the underlying disks, stretched clusters require an underlying infrastructure capable of delivering high I/Os with low latency. Dell Technologies recommends all-flash configurations for stretched cluster deployments.

All-flash configurations do not have a cache tier. The following table shows the difference in storage efficiency for a two-way and three-way mirror created on a single site and stretched cluster environment:

Table 7. Storage efficiency differences

Single site

Two-Way Mirror

50%

Stretched cluster 25%

Three-Way Mirror

33%

16.5%

Test-SRTopology

This cmdlet validates a potential replication partnership. The cmdlet:

● Measures bandwidth and round trip latency

● Estimates initial sync time

● Verifies that source and destination volumes exist

● Verifies that there is sufficient physical memory to run replication

NOTE: The file server feature has to be enabled on the nodes to run this cmdlet.

Keep the generated report for future reference.

For more information about this cmdlet, see Test-SRTopology .

Creating a stretched cluster 19

advertisement

Was this manual useful for you? Yes No
Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Related manuals

advertisement

Table of contents