Design Considerations for a Virtualized HA Environment

iSCSI SAN Storage Software / Features / StarWind’s Active-Active HA storage architecture / Design Considerations for a Virtualized HA Environment

Design Considerations for a Virtualized HA Environment

Planning any sort of high availability system requires careful consideration of the functional components of your system, both from a software and hardware standpoint. Because most applications are backed by a relational database, SQL Server high availability or DB2 high availability planning involves a deep understanding of active-active and active-passive failover configurations as well as some experience in the planning and deployment of fault-tolerant storage subsystems. Typically, these utilize RAID and/or some form of a storage area network (SAN) to ensure the system remains up – and no data is lost – in the event of a single hard drive or network failure.

Application servers are often deployed within high availability clusters using some combination of software or hardware load balancing. Finally, network availability should be considered, and careful design consideration should take place to eliminate single points of failure at the network interface card (NIC), switch, and router levels.


High availability architecture based on virtualization software requires the special set of design considerations to ensure that the virtualized environment scales and remains available as multiple nodes are added to the highly available cluster. Virtualized environments based on StarWind iSCSI SAN software provide an unprecedented degree of flexibility to system architects by allowing individual virtual machines (VMs) to be sized and scaled on the hardware based on their individual needs. Therefore, the first consideration would be to determine which VMs need to participate in the high availability system and which do not.

The second design consideration involves properly estimating the server resources that each individual VM will require at runtime, in order to be sure that all VMs have the CPU, memory, and storage resources that they actually need, without overconsumption.

In addition to the considerations mentioned above, many design issues are identical between a normal physical environment and virtualized ones. For example, in both models, high availability storage solutions are required so that failures in the underlying storage subsystem will not bring the system down. Improper storage system design is the most common cause of high availability system failures so it is important that this be properly handled at the design phase.

The final design consideration for any virtualized HA environment involves testing and StarWind based one is not an exception. Take care to plan how the HA environment will be tested during the implementation phase, and also when the environment is in production. It is possible to deploy high availability virtualization solutions, assuming the proper considerations are taken during the design phase of your project.