Implementing vSphere Metro Storage Cluster using HPE XP High Availability

Products

• VMware vSphere ESXi

Issue/Introduction

This article provides information about deploying a vSphere Metro Storage Cluster (vMSC) across two data centers using HPE XP7 or XP8 Storage Arrays.

Resolution

What is vMSC?

vSphere Metro Storage Cluster (vMSC) is a tested and supported configuration for stretched storage cluster architectures.  A vMSC configuration is designed to maintain data availability beyond a single physical or logical site. All supported storage devices are listed on either the VMware Storage Compatibility Guide or the Partner Verified and Supported Products (PVSP) listings.

What is HPE XP High Availability?

XP High Availability is the HPE XP7 and XP8 Storage Array technology for active-active stretch storage clustering allowing for unprecedented uptime using virtualized replicated volumes at metro distances.  XP High Availability data volumes are synchronously replicated across storage arrays bi-directionally and virtualized to the host.  By using Asymmetric Logical Unit Access (ALUA) capabilities that allow paths to a SCSI device to have different characteristics, hosts connect to volumes on one array via active optimized paths and connect to the replicated array via active non-optimized paths.  Volumes are read/write enabled on both arrays at the same time, enabling virtual machine migration and failover/failback between sites with zero downtime.

Configuration Requirements

The following requirements must be satisfied to support a vMSC configuration with HPE XP7 or XP8 Storage.

• VMware ESXi 6.0U2 or higher, Metro Storage Cluster configured for uniform (recommended) or non-uniform host access per VMware requirements and best practices
• HPE XP7 or XP8 Storage arrays configured using XP High Availability per HPE XP High Availability User Guide for the relevant array
  • Maximum round trip latency on the storage network between sites should not exceed 20ms RTT
• A fiber channel or iSCSI based quorum device hosted on a supported storage array or host.
  • Quorum Device must be used at a third site.  iSCSI disks as well as virtualized storage arrays are supported
• vSphere vCenter Server connected to ESXi hosts in both datacenters
• FC-switched SAN for host-array connectivity, remote copy path connectivity 
• FC-switched SAN for array-quorum connectivity if a FC based quorum device is used or a IP based network for array-quorum connectivity if an iSCSI based quorum is used

Solution Overview

VMware vSphere Metro Storage Cluster using HPE XP Storage using XP High Availability provides an ideal solution for maximizing availability by clustering servers and storage at data centers within metro distances.  XP High Availability provides replicated volumes across storage arrays that are geographically distributed.  This provides an environment in which virtual machines can be migrated transparently.

The VMware Metro Storage Cluster with HPE XP Storage solution is composed of the following parts:

• vSphere 6.0U2, 6.5 or 6.7 ESXi hosts
  • The hosts running virtual machines at both data centers
  • Native VMware multi-pathing
  • Host Access via uniform (recommended) or non-uniform access vMSC Configurations
• HPE XP7 or XP8 Storage
• HPE XP High Availability
• Quorum Device
  • A volume presented from any supported HPE XP External Storage array or an iSCSI host, used to prevent split-brain conditions
• Fiber Channel Switches
• Network Switches

In a uniform access configuration, host fiber channel connections are cross-connected to the other data centers storage in addition to the local storage array.  By utilizing the ALUA (Asymmetric Logical Unit Access) feature of XP High Availability, the hosts will favor access to the local site but can utilizes paths to the other site in the event of an issue.  Both sites volumes are read/writeable under normal operating conditions until such a time that some part of the solution fails.  At the point of failure, one of the XP Storage Arrays will be selected to remain in a read/writeable state while the other will become read/write blocked automatically.

For example, in the case of an array failure at site A:

1. The site B array will notice a loss of connection and attempt to gain a lock on quorum
2. Upon receiving a lock, the site B array will continue read/write operations
3. Hosts will continue working utilizing cross-site (for hosts at site A) or local paths to the storage on site B

Note: Storage failover is an automatic process, recovery is a manual process.

This diagram depicts a high-level overview:


Tested scenarios
The following table outlines some supported failure scenarios when using uniform host access configuration of XP7/XP8 High Availability and VMware vSphere:

Scenario	HPE XP8 Storage Behavior	VMware HA Behavior
Single Array-Host Path Failure	Hosts use alternate paths to maintain volume access	No effect
Single Remote Copy Path Failure	Arrays use alternate remote copy paths to continue replication	No effect
Complete Remote Copy Path Failure	Replication is suspended, last array to perform I/O remains active, other array is read/write blocked	No effect
Quorum Failure	Replication continues	No effect
Quorum Failure, then Complete Remote Copy Path Failure	Replication is suspended, array hosting the Primary High Availability Pair volume remains active, other array is read/write blocked	No effect
Single Site Storage Array Failure	Replication is suspended, volumes on surviving site remain active	No effect
Complete Site Failure	Replication is suspended, volumes on surviving site remain active	Virtual Machines failover to surviving site hosts