With all of the recent push to the cloud by various vendors it’s easy to forget that there are really two ways of implementing cloud computing. For sure, the big cloud vendors push is for businesses to move to public cloud computing. However, many organizations have leaned more toward the private cloud.
The public cloud and the private cloud offer similar promises: reduced operation expenses, greater elasticity of resources, and self-service management. The big difference is that using the public cloud requires leasing resources from some other vendor’s web-based infrastructure. In contrast, you implement the private cloud as a management layer on top of your own internal infrastructure. Not surprisingly, most IT professionals are far fonder of the idea of the private cloud than they are the public cloud because the private cloud leverages your existing infrastructure and its keeps vital aspects of your infrastructure like performance, availability and security under the control of your own company.
But if you think about it does the private cloud really work for SQL Server? SQL Server is not like IIS where workloads can be balanced between multiple servers. In other words, you can’t add an additional SQL Server instances to your cluster and get any performance benefits. For SQL Server, clusters only offer enhanced availability—not enhanced performance. SQL Server users are connected to a given database and that database has a given configuration and properties. While that may be the case that doesn’t mean that you can dynamically alter the available computing power for a SQL Server instance.
Virtualization is the key to implementing SQL Server in the private cloud. SQL Server instances running in VMs can be dynamically live migrated to hosts with greater resource availability and computing power. Further, if the SQL Server instance is running in a VM and the host has the available physical capacity you can dynamically increase the number of virtual CPUs and RAM and available to the SQL Server instance. Hyper-V and vSphere both support hot-add RAM and CPU. In addition, SQL Server has the ability to be reconfigured to accept hot-add RAM and CPU without requiring any downtime. Live migration, hot-add RAM and CPU let give SQL Server the elasticity it needs to be a fully functional component of the private cloud.