Four considerations for server Application Virtualization

As terminal virtualization and Application Virtualization grow, IT managers must recognize that server requirements can change dramatically because of the different virtualization technologies that are selected for application delivery.

Traditional non-virtualized applications are delivered, applications need to be installed on a physical server, and end users access applications over the LAN. Because this delivery model consumes only 5% to 10% of the server's computing resources, few people pay attention to the overall configuration of the server.

As long as the server meets the computing needs of the application, administrators rarely pay attention to application-delivered resource consumption until there are obvious performance issues associated with the overall server configuration. For example, a large number of users may degrade network performance, and administrators may bind multiple network adapters to resolve the problem.

The application of terminal virtualization has completely changed this approach. Virtualization improves server usage and allows physical servers to handle more concurrent tasks. But administrators face the challenge of planning. For an application delivery server, sufficient computational resources must be allocated to support end users or terminal nodes, and continuous management is required to avoid resource depletion and poor performance. In addition, virtualized servers must provide enough flexibility to minimize downtime and user disruption. Virtualization has a higher demand for IT specialists to coordinate and balance resources.

Consider applying virtualized servers

Applying a virtualization server allows users to access independent instances of an application that is actually installed on the server. But application virtualization has a high demand for I/O performance. For example, the choice of storage location may have a significant impact on application virtualization.

"Any time you deploy a storage area network, the first thing to consider is I/O issues," said Lan Parker, a senior Web service administrator at global information resources. "We've been keeping a close eye on the development of flash drives lately. ”

Disk I/O performance is important to the storage array, but I/O problems may also come from the network. For example, Ethernet based Sans, such as iSCSI or FCoE, can create network bandwidth bottlenecks, so storing the storage on a separate application server simplifies the potential network congestion problem. Network I/O performance also responds to bandwidth-intensive application flows, and bottlenecks arise when the application is paid to the terminal on demand, which is different from running the application completely on the central server.

Parker says memory is not a significant problem for application virtualization because modern servers with 64-bit operating systems can easily support hundreds of gigabytes or even terabytes of memory.

Consider desktop instance (VDI) servers

A virtual Desktop interface (VDI) server hosts an entire desktop instance on a central server, exchanging only user input and audio/video output information via a simple terminal device on the user's side, which is generally referred to as "thin client" or "0 client". Servers hosting an entire desktop instance are more sensitive to resource overhead than application virtualization, so you should consider upgrading to more powerful CPUs, memory, and storage I/O. Once the desktop instance completes loading and starts running, the requirements for network I/O are reduced to very low.

Local storage can effectively improve VDI performance, but sans are a more popular storage platform because they also provide a single point of management capability.

"If you are planning to deploy VDI on a SAN, that bandwidth becomes very important," Parker said. "Then there is memory and disk throughput, because you'll find a lot of paging and other activity." ”

The VDI instance on the delivery server will continue to grow, and the disk subsystem must meet the growing performance requirements. For example, Parker found that when the VDI instance started the system, the primary activity on the primary storage was write disk, and the disk array method for improving write performance should be used, while the poorly written RAID5 array might affect the performance of the storage subsystem.

Personal desktop instances can also improve performance by adding the number of server CPU cores. Therefore, selecting a CPU with multiple cores when configuring the server can elevate the server's VDI hosting capability.

"Select a large server to achieve, the more core CPU than the core of the faster," Bob plankers,the Lone SysAdmin's technical advisor and blog columnist.

Consider the application of a single resource requirement

The application for virtualization may have additional requirements for some configuration of the server. For example, a medical imaging application designed to handle a large number of files may have higher requirements for memory and storage I/O. Peak performance may occur after applying virtualization and delivering services to multiple users. such as Liquidware Labs Inc. 's tool stratusphere can help administrators diagnose the resource overhead required before migrating to VDI deployments.

Similarly, consider the growth of other virtualization applications, such as computer-aided design images and rendering tools. Integrated vision technology, similar to RemoteFX in Windows Server 2008 R2 SP1 and ' HDX in Citrix Bae Inc.s 4, can be computed using a strong video card on the Terminal server. This is another step forward for application delivery, but IT administrators need to weigh the impact of this technological advance on the server infrastructure. "What type of blade server can support the current mainstream PCIe interface type graphics card with additional consideration for space and cabinet occupancy issues," Parker said.

However, there is no niche requirement for the current major business applications, and virtualization applications will only have some impact on the server. "That's what we're focused on, but it doesn't have much impact on how we choose servers," Parker said.

Consider cluster and application delivery elasticity

Providing key servers for enterprise users to deliver critical applications often requires sufficient flexibility to ensure availability. The server itself provides some flexibility, such as a onboard RAID controller for managing local disk storage and redundant power supplies. Traditional server cluster consolidation tools or recently developed virtual machine redundancy technologies can protect the load from unplanned downtime. This is a more important aspect of application virtualization because any application terminal can affect a large number of users.

(Responsible editor: The good of the Legacy)