Scanning virtual server infrastructure has become necessary in the face of increase in server workloads. It needs to be analyzed not only for expected results but also because such measures provide scope for improvement in the virtualized systems.
Routine problems in a virtual server environment: Just as in a physical server environment, when a problem occurs, the administrator inspects a switch port or a host bus adapter, similarly, in a virtual environment a logical fault finding process would be followed to find the root cause of an issue. Each virtual component (such as a network interface card, CPU or memory) can be monitored to determine its level of functionality and utilization. It is necessary to also monitor the underlying physical server infrastructure, and this is best done through systems management tools.
Physical and virtual server resources should be managed in a cohesive manner using tools that allow for workloads to be delivered to the correct platform. Ideally, a comprehensive management suite that monitors the entire virtual infrastructure as well as the virtual servers should be implemented.
Handling host failure in a virtual server environment: Server virtualization has allowed organizations to increase efficiency by making better use of shared server resources – a single physical server can be resourceful to multiple virtual server environments, allowing organizations to extract maximum value out of their hardware assets. However, it is necessary to understand the impact it can cause should the physical server fail. Hence, strategies need to be put in place to firstly make the hosts more tolerant to failure through component level redundancy, and secondly, to allow for server workloads to be transferred to another host should there be a total failure.
Massive bloat in server count: Prior to virtualization, organizations could not provision new workloads to the point where the number of logical server instances increases beyond a few numbers. Each logical server instance consumes resources, particularly storage resources. Each instance needs to be managed individually to ensure it has applied the correct service packs, hotfixes and security patches. This can strain the administrative staff and processes who had thought that server virtualization would reduce the overheads associated with administration tasks.
One should deploy a solution which allows a single virtual hard disk to be provisioned to many systems — be they physical or virtual. This single instance model reduces the storage overhead associated with large number of virtual machines because they can all be serviced from a single virtual hard disk. It also focuses administrative effort onto the single instance of the virtual hard drive rather than the virtual servers. No matter how many virtual machines are being serviced, the administrator only needs to update and maintain the virtual hard disk – any patches or services packs are applied only once then distributed to all the virtual machines at the next reboot.About the author: Souma Das is the area vice president for Citrix, India Subcontinent.
(As told to Jasmine Desai.)