How to create a Microsoft SQL Server failover cluster in the Google Cloud


While the cloud continues to grow in popularity for many, if not most, enterprise applications, IT departments remain reluctant to trust public clouds, including the Google Cloud Platform, for business-critical Microsoft SQL Server applications. Their primary concern is the increased risk and complexity of running Microsoft SQL Server, and optionally Microsoft’s Windows Server Failover Clustering (WSFC), in environments where high-availability clustering configurations are often less robust and more expensive than they are in a private enterprise cloud.

Among the issues for SQL Server is the need to use shared storage for Always On Failover Cluster Instances. This is a problem for two reasons. First, shared storage in the form of a storage area network (SAN) is not available in the Google Cloud Platform. Second, shared anything creates a single point of failure when not also made fully redundant.

One way to overcome such limitations involves implementing third-party failover clustering software within the Google Cloud. When designed to support SQL Server, the failover cluster creates a high-performance, high-availability environment that is easy to manage.

Limitations on SQL Server high availability in the Google Cloud

Like other cloud services, the Google Cloud Platform offers application failover capabilities for many applications requiring high availability. These services work by replicating data from the master instance to one or more replica instances in different zones, which are normally located in different data centers to protect against localized disasters. In the event of a failure in the application’s master instance, the high-availability provisions fail over to a replica instance.

to create the SAN-less failover cluster. SIOS DataKeeper uses block-level replication to ensure that the locally attached storage on each SQL instance remains synchronized. SIOS DataKeeper also integrates with Windows Server Failover Clustering through its own storage class resource called a DataKeeper Volume, which takes the place of the physical disk resource. Within the cluster, the DataKeeper volume appears to be a physical disk, but instead of controlling SCSI reservations, it controls the mirror direction, thereby ensuring that only the active server writes to the disk and that passive servers receive all the changes, either synchronously or asynchronously.

The architecture for a SAN-less failover cluster configured as a virtual private cloud (VPC) is shown in the diagram below. There are two SQL Server nodes in different zones (1-A and 1-B) in the US-Central region. The File Share Witness, which is needed to configure a quorum in Windows clusters, is performed by the domain controller in US-Central zone 1-C. Keeping each instance of the quorum in a different zone eliminates the possibility of losing more than one vote if an entire zone goes offline. Also shown are the VPC’s node names, IP addresses, and subnet values that will be configured in the steps that follow.


Shown here is the final configuration of the SAN-less failover cluster for the Google virtual private cloud.

Failover clusters are created in the Google Cloud in much the same way they are created in a private cloud, but with one important difference: You can’t use a single subnet with a single virtual IP address that moves among cluster nodes, owing to GCP’s lack of support for gratuitous ARP (address resolution protocol). It is possible to get around this limitation by following procedures outlined here that place each node in a different subnet, and then create host-specific routes for the cluster IP addresses. Note that the discussion assumes the reader is familiar with both the Google Cloud and IP address subnetting.

Step #1: Create the VPC network and failover cluster instances

In this example the VPC, named WSFCNET, has three subnets: WSFCSUBNET1, WSFCSUBNET2, and WSFCSUBNET3 in three different zones (,, and for SQL1, SQL2, and DC1, respectively. Using the Google Cloud Platform console, each node will initially be given a static IP address with forwarding in the /24 subnet; however, we will change all of the SQL nodes to a /16 subnet mask to support host-specific routing. This step also involves creating firewall rules to permit communications among the VPC nodes.

Here is the command for creating WSFCSUBNET1:

.” While some of the screens and settings are specific to SIOS DataKeeper, the configuration requirements are fully applicable to any SQL Server failover cluster solution.

David Bermingham is a high availability expert and eight-year Microsoft MVP, six years as a Cluster MVP and two years as a Cloud and Datacenter Management MVP. David’s work as director at has him focused on evangelizing Microsoft high availability and disaster recovery solutions and providing hands-on support, training and professional services for cluster implementations. Email David at  and learn more at and . 

New Tech Forum provides a venue to explore and discuss emerging enterprise technology in unprecedented depth and breadth. The selection is subjective, based on our pick of the technologies we believe to be important and of greatest interest to InfoWorld readers. InfoWorld does not accept marketing collateral for publication and reserves the right to edit all contributed content. Send all inquiries to .