Using HP Serviceguard for Linux with VMware virtual machines - Technical white paper

ManualsBrandsHP ManualsSoftwareHP Serviceguard for Linux License Kit

About this paper

Virtual machine technology is a powerful capability that can reduce costs and power usage while improving utilization of

resources. HP is also applying virtualization to other aspects of the data center and uniting virtual and physical

resources to create an environment suitable for deploying mission-critical applications.

HP Serviceguard for Linux is certified for deployment on Linux virtual machines created on VMware ESX server on

industry-standard HP ProLiant servers.

You can configure Serviceguard clusters consisting of physical and virtual machines

running these certified Linux distributions. For the complete list of hardware and versions of VMware certified and

supported by Serviceguard, refer to: hp.com/info/sglx

→HP Serviceguard for Linux Certification Matrix.

This paper provides information on configuring a Serviceguard for Linux cluster that includes virtual machines; it also

makes recommendations for eliminating single points of failure and provides pointers to other useful documents.

Introduction

Virtual machines on VMware are being deployed increasingly for server consolidation and flexibility. Virtual machine

technology allows one physical server to simulate multiple servers, each concurrently running its own OS. In virtual

machine technology, the virtualization layer also known as hypervisor

abstracts the physical resources so that each

instance of an OS appears to have its own NIC, processor, disk, and memory, when in fact they are virtual instances. This

allows you to replace a number of existing physical servers with just one, but at the cost of greater exposure to

failures—with dedicated physical servers any failures would only affect the applications running on it, but in virtualized

environments a physical server failure results in a number of virtual servers failing along with all the applications

running on them.

VMware has a high-availability (HA) clustering product called VMware HA. It can provide some degree of protection from

failures, but it has limitations. VMware HA uses a simple model that detects only physical server failures. When it detects

those failures it restarts the virtual machines from the failed server on other servers running VMware. Serviceguard for

Linux that runs in the virtual machines provides a significant level of extra protection. Specifically, Serviceguard for

Linux fails over an application when any of a large number of failures occurs, including:

• A failure of the application

• A failure of networking required by the application

• Failure of storage

• An OS “hang” or failure of the virtual machine itself

• Failure of the physical machine

There are other advantages beyond the increased failure protection.

• Serviceguard for Linux failover is faster than VMware HA. Serviceguard for Linux restarts just the application, while

VMware HA requires the virtual machine’s OS to boot on the failover target before restarting the application

• Serviceguard for Linux rolling upgrade feature allows for less planned downtime

– With Serviceguard for Linux, an application (packages) can be moved off a virtual machine and restarted on another

node in the cluster. The “empty” server can then have its OS or applications upgraded while those applications are

still available to users since they are running on other nodes

Serviceguard for Linux, combined with VMware, can provide a lower-cost method of protecting applications that would

not normally be run on a virtual machine. Applications can be failed over between physical servers and virtual servers on

the same or different physical hosts. Cost savings and some additional reliability can be realized by combining physical

and virtual machines in a cluster. In some environments, this configuration provides better utilization of data center

resources. Users often configure clusters with one node as a “backup” node. When there are multiple clusters, the

number of physical servers can be reduced as there is no need for dedicated “backup” hardware for each cluster. One

physical system hosts backup nodes for multiple clusters as virtual machines.

For latest details, see hp.com/go/vmware, http://www.vmware.com/solutions/partners/alliances/hp-resources.html

Hypervisor often refers to a layer that resides directly on server hardware, but terms are not used consistently across the industry.