Virtual Server management experience and skills

Under the premise of correct implementation and management, virtualization means cost saving. Here there are savings in server hardware, support and management, there are also simpler server deployment and further reduced energy consumption. To make these benefits a reality, the management of the virtualization layer of this structure must be independent of applications or physical servers.

Integrity

Although you do not want a virtual server with the same requirements to be located on the same physical server, you may want to place some physical servers at least physically close to the network.

Cause: in order to reduce the number of virtual servers to be switched between physical servers, it needs to increase. If a group of physical servers share network connections and other resources, the time it takes to switch a virtual server from one physical server to another will be greatly reduced.

"If you have multiple servers in a rack, you can shut down the Virtual Machine on one server and open it on another server on the same switch," Anderson said, "The time spent is about the same as the time used to copy the hard disk."

On the other hand, he said that if the server you are converting is located at the other end of the data center, it needs to be inserted into different subnets of different switches, the configuration time is increased by 5 or 10 minutes. Or if there are other restrictions, it may take more time.

Of course, there are limits on proximity, whether physically or on the network. For example, you do not want servers to overload network connections, and you want physical servers to be separated by a distance or in a different state. for disaster recovery. This is a balanced action.

Tracking hardware

This non-linearity complicate the management of virtual servers. Administrators need to closely monitor resource requirements on different physical servers.

This is different from the requirements reported by the operating system on the virtual machine. As the configuration files of application resources change, administrators are advised to pay close attention to the trend to avoid sudden resource shortage.

Moreover, tracking of physical hardware must be fine-grained, because virtualized applications rely on different types of resources provided by the server for maintenance, resulting in different loads. Because different virtualization applications have different requirements, such as memory, processor cycle, and I/O bandwidth, they need to be tracked separately.

What you cannot do is monitor the operating system on the Virtual Machine to track the physical resources of the server. You must track resources on a physical server at the lower layer of the hypervisor and perform operations from different virtual machines. You may still need to track the resource utilization reported by the VM, but it is an independent problem, it is done for performance analysis and so on. (Note again that you are using hypervisor virtualization software such as Vmware or Xen ).

According to Anderson, storage bandwidth is usually the most critical resource on physical servers, just because it is a hard limit. Unlike other aspects, from storage to processing cycle, you cannot increase the storage bandwidth built into the server. When it is saturated, You need to migrate some virtual machines to a new physical server.

Bandwidth of all types is a specific problem because it may be the problem. Not only are almost all applications using bandwidth, but their load also tends to be diversified. If two or more applications meet the maximum memory requirements at the same time, you may encounter problems that are difficult to track.

This brings about a new level of complexity, or at least a new observation point in the virtualized data center.

When physical servers that support virtual machines consume resources, a large number of applications are affected.

Monitoring

Ensuring resource availability is another reason for the close monitoring of virtual/physical server stacks. Another reason is that hardware failure affects more than one application.

Because monitoring is so important, mongoozzo virtualization software such as Vmware and Swsoft includes tools that can help track physical and virtual resources.

Because the load changes with time, monitoring is a continuous process. You cannot set monitoring and forget to monitor it. In fact, you must pay more attention to monitoring because each physical server is supporting several virtual servers, and if a physical server begins to exhaust resources, all the virtual servers will be affected.

Related Articles]

• Microsoft relaxed authorization for 41 server programs for Virtualization

• Use cross-platform virtual security tools to enhance virtual security

• Ten misunderstandings of server virtualization technology