|
In recent years, server virtualization has evolved from a technology with significant usage in development, training, and test environments to one that also has a viable place in the data center. With these changes, it’s no coincidence that information technology (IT) staffers no longer hum to the tune of Olivia Newton John’s “Let’s Get Physical.” Space and power limitations in the data center have fueled a large consolidation movement, with server virtualization and clustering at the forefront.
While virtualization allows organizations to run multiple unique operating systems (OSs) on the same physical host simultaneously, it also offers other benefits in high availability and system portability. Naturally, the benefits come with tradeoffs including potential performance degradation, as well as challenges associated with network and storage area network (SAN) integration, backup, and system management.
The tradeoffs for server virtualization solutions often vary both by virtualization architecture and product vendor. Host-based server virtualization (fueled by market leader VMware) provides excellent options for system consolidation, portability, storage integration, and automated failover. Evolving host-based virtualization architectures such as paravirtualization and hardware-assisted virtualization have led to significant performance improvements. OS virtualization offers superior consolidation ratios, yet does not provide the same flexibility in system portability offered by host-based server virtualization.
Ultimately, deciding on a virtualization platform boils down to how well a product can take advantage of virtualization-enabled hardware as well as integrate with the existing network infrastructure. Support for 802.1Q virtual local area network (VLAN) trunking and N_Port ID Virtualization (NPIV) SAN integration is dependent on both the selected virtualization application and the hardware, whether existing or planned, that will be used to connect virtualized hosts to an organization’s resources.
Virtualization adoption in the data center offers several benefits: efficient hardware resource utilization, server portability, and high availability for any system (including applications that do not natively support clustering). To realize those benefits, careful evaluation of prospective virtualization products is paramount. With numerous moving parts (Fibre Channel or Internet Small Computer Systems Interface [iSCSI] SAN, network infrastructure, and data protection and management applications) affected by any conversion to virtualized resources, awareness of each virtualization technology’s limitations is essential. There is little room for error when it comes to managing data center resources. Understanding where each virtualization technology is best suited in the data center allows organizations to realize the benefits of virtualization without falling victim to its weaknesses.
Analysis
Server virtualization has quickly jumped from a niche to a mainstream technology over the last couple of years. Organizations faced with limited power and space in the data center have looked to new technologies to enable data center expansion while reducing hardware and power requirements at the same time. Today, there are two primary approaches to solving the problems of growth while consolidating: clustering and server virtualization.
Many organizations have turned to clustering as a means to consolidate high-performance applications while also providing high availability and failover support. While clustering is proven in the data center, it has its limitations, like all other technologies. A primary restriction of clustered solutions is application and service support. Deployment complexity and management requirements also preclude clustering solutions from being viewed as viable for some applications and services. Given that many third-party applications do not support clustering, nor have a proprietary cluster model, clustering in itself is often not enough to solve every consolidation problem.
Server virtualization provides a means to consolidate multiple servers onto one or more physical systems. Virtualization technology may also allow the relocation of servers to systems with completely different hardware without any downtime. With server virtualization, each virtualized host becomes a unique virtual machine (VM), with major elements of the VM’s hardware emulated. For example, motherboard emulation allows a VM to be copied or moved to a system with a different physical motherboard than the VM’s original host. To the operating system (OS) running inside the VM, no difference exists between the hardware seen on one physical host versus another.
On the surface, server virtualization holds tremendous potential for consolidation. Advanced features such as automatic failover, dynamic relocation, load balancing, and consolidated backup have pushed many organizations to embrace and use this technology in production.
|
