You can save money and gain tremendous IT agility, flexibility and efficiency in your data center by moving your applications to virtual server and cloud environments. However, there is significant confusion in enterprise IT management about how to meet the exacting availability requirements of business critical applications in these environments.

IT managers have a wide range of questions: Do you need high availability (HA) and disaster recovery (DR) protection in a public cloud? Are replication solutions sufficient? Can you provide high availability and disaster protection that doesn’t limit or offset the cost saving and flexibility benefits of virtualized environments? While HA and DR in a physical server environment is a straightforward matter of setting up a shared storage cluster, in virtual server and cloud environments, the best practices for HA and DR are not as clear.

Traditional HA and DR Protection

The traditional shared storage cluster is the preferred solution for achieving HA for business critical applications. When protecting applications based on SQL Server, IT staff often use Windows Server Failover Clustering (WSFC) to configure a cluster of two or more servers that share the same storage, usually via a SAN or NAS device. The application runs actively on one server and WSFC software monitors the server and application. If WSFC detects a failure, it orchestrates a recovery of all required application resources on another cluster node.

To achieve DR and protect applications from site failure today, IT typically uses a variety of replication or failover technologies and they require multiple products, scripts and manual steps to allow recovery at a remote site. Alternatively, more costly enterprise-grade SAN solutions are used to implement geo-clustering across geographically separate sites.

Challenges of Protecting Applications in Virtual Server and Cloud Environments

When moving to virtualization and cloud platforms, there are a number of application, database and hypervisor or cloud specific approaches to use for availability. Examples of database level solutions include SQL Server AlwaysOn Availability Groups (AOAG). In the cloud, regional isolation of workloads is an option typically provided as an add-on service to use for assembling an HA/DR configuration. However, there are limitations in these solutions that make achieving cost-effective and reliable availability challenging.

For instance, AOAG is a SQL database-only protection solution and does not replicate the entire storage volume, which may contain other application data. It uses log-shipping and can add significant CPU overhead slowing performance.

In cloud environments, most providers offer separate computing resources for local or site redundancy. Providers also offer data replication services between zones but do not include any application monitoring or failover capabilities. You’ll need to use other technologies to fashion a complete solution to achieve HA or DR.

In contrast to these approaches, traditional shared storage clustering is a comprehensive, full stack HA solution that protects the application across the OS and all lower operating software and hardware layers. The problem is that traditional clusters require shared storage limiting their usefulness in virtualization and cloud environments.

For instance, you can use shared storage clustering in the guest VM, but to do so you have to use Raw Device Mapping (RDM), which adds complexity and limits the flexibility. In cloud environments, shared storage isn’t even an option.

So, if shared storage clustering is the best solution for tier-1 application availability, is there a way to provide availability and disaster protection using clustering without shared storage?

SANLess Clustering for HA and DR Business Critical Applications

A simple way to gain the advantages of clustering in virtual server and cloud environments without the limitations of shared storage is to add SANLess clustering software to a Windows Server Failover cluster. This software allows you to create an HA cluster without the need for shared storage. It uses efficient, block level host-based replication to synchronize local storage on each cluster node so that it appears to WSFC as shared storage. It can be used in virtual server (any hypervisor), cloud, hybrid cloud, and in physical server environments using high performance (SSD) storage.

In some environments, you can configure SANLess clusters in the guest VM and use VMDK storage, eliminating the need for RDM. SANLess clusters give you both full high availability protection for your applications and data and flexibility. Using SANLess clusters allows you to use low cost local storage or high performance server-side flash to cluster. This gives you complete VM portability, live migration, and failover protection with very high performance.

To take advantage of the cost savings, flexibility and efficiency benefits of virtual server and the cloud you need a sound strategy for protecting important application environments from downtime and data loss. While the redundancy offered in cloud and some virtual server environments provides some level of protection, it is not sufficient for business critical applications. SANLess clusters provide an easy, cost-efficient way to protect applications in these environments where traditional shared storage clusters may not be practical or even possible.

About the Author:

Jerry Melnick is the chief operating officer of SIOS Technology Corp. Melnick (jmelnick@us.sios.com) is responsible for defining corporate strategy and operations at SIOS Technology Corp. (www.us.sios.com), maker of SIOS SAN and #SANLess cluster software. He has more than 25 years of experience in the enterprise and high availability software industries. He holds a bachelor of science degree from Beloit College with graduate work in computer engineering and computer science at Boston University.

Read this article at drj.com