Rajen Naicker, Overland Product Specialist at Drive Control Corporation (DCC), tackles the challenges around the much-discussed virtualization.
Virtualization is becoming a more common practice in the ICT space, with the virtualization of servers and the dividing of one physical server into two or more virtual server machines being the most common form of this. Virtual Machines, or VMs, which run as several dumb terminals off one server machine, are also becoming far more common.
In fact, according to the Worldwide Enterprise Server 2008 Top 10 Predictions report by IDC, 39% of new servers deployed over the next five years will be virtual servers.
However, along with the benefits of virtualization, such as lower power consumption and cooling requirements, easier maintenance and reduced physical storage space, the virtual world also presents a number of challenges, not least among them how this phenomenon increases the complexity of the backup of data, business continuity and Disaster Recovery plans.
The effect of virtualization. Virtualized servers affect data storage and protection in several ways. The first of these is due to the consolidation of inputs and outputs. Because many virtual machines in this type of environment are consolidated onto one physical server, the input/output demands on the consolidated server are cumulative, so any storage device used needs to be able to support higher input/output per second as well as higher rates of data transfer.
Added to this is the need to support a larger number of inputs and outputs per second, as well as more data streams. Within a virtualized environment, more applications must run simultaneously, so it is vital to use storage that supports multi-threading and multi-paths to ensure consistently high levels of performance as well as to avoid bottlenecks.
There is also the possibility of resource contention between servers and storage, as many storage applications, such as backup, are very resource intensive and running these concurrently across multiple virtual machines on the same physical server can cause a fight for necessary resources, slowing the system down and causing undesired effects.
As a result, virtual server environments are driving a growth in Server Area Network (SAN) connectivity via iSCSI and/or Fibre Channel, in order to be able to support VM mobility, aggregated inputs and outputs and shared SAN device, further complicating issues such as storage and backup.
Data protection needs careful planning. Adequate and well thought out planning of data protection is critical in a virtual world due to the fact that Virtual Machines are consolidated onto fewer physical servers. Backup is resource intensive as a result in terms of CPU, memory, network and storage.
It is important to remember that many of the best practices and policies that apply to traditional physical server data protection are still valid and have a place. However when conducting VM data protection, it is vital to ensure that any backup plan anticipates the resource impact of having several Virtual Machines on one server.
Backup can no longer be taken as a standalone application that affects only a single server or application, as each VM backup can have an effect on other Virtual Machines sharing the same virtual server. Data protection planning adds a new layer of complexity that must be carefully considered when a virtual server strategy is implemented, in terms of the impact on other VMs as well as the physical resources these need.
Addressing virtualized server issues. Virtualized server environments are complex to back up and raise many issues, such as whether to back up the core server platform, the guest operating system platform, the VM data or the full VM and so on.
Luckily these questions and many others can be addressed by three basic VM data protection models, each of which has its own merits depending on the needs of the environment.
The first of these is Virtual Server Console Backup, which provides basic file and image backup solutions when the backup is loaded onto a virtual server console. The second model is Virtual Machine-Based, which entails the backup software and agents being loaded on the VMs as if they were actual physical servers. Finally there is Group VM Consolidated Backup, which leverages virtual server-based tools for consolidated or group backups of Virtual Machines.
To conclude, building an efficient data protection strategy for Virtual Machines and virtualized server environments necessitates a thorough process and much consideration.
In order to ensure that a solid plan is built and that the overall impact of VM backup is understood, it is necessary to get expert advice from an experienced data protection provider. This will ensure that an organization has the best virtualization strategy is in place to ensure optimal performance and data protection.