Two common traps of server virtualization and Their Countermeasures

Virtualization is regarded as a panacea for solving many enterprise IT problems. From enhanced application availability to simplified disaster recovery, to reduced infrastructure and costs, virtualization seems to provide all the answers. Virtualization also seems to provide simplified IT management and even "greener" computing solutions.

However, to maximize the benefits of server virtualization, other elements of infrastructure, especially storage, are important to make up for the shortcomings of the virtualization environment. Otherwise, many errors will occur. The application will be unexpectedly slow as crawling. It was originally an alternative to computing to reduce costs, but it took a lot of investment to achieve all its functions. Using virtualization to improve the normal running time of applications and servers suddenly exposes pain points in other aspects of the IT infrastructure. The following are the two most common traps for enterprises to implement server virtualization.

Trap 1: select the wrong storage platform

One of the main advantages of server virtualization is the ability to move the customer applications in use between the management programs of different servers. Whether it is for orchestration planning, load balancing, or disaster recovery, hardware independence is one of the main driving factors supporting any virtualization implementation. However, if your storage is associated with the specific server hardware, mobile applications will become more complex and even unavailable.

Network attached storage is often used as a method to simplify the storage configuration of virtual servers. The setting of additional network storage capacity is very simple, and the expansion of the capacity does not require the participation of management programs. Unfortunately, network attached storage has performance weaknesses. Many applications, such as Microsoft Exchange, cannot run with network attached storage. For these reasons, most virtualization vendors will recommend the use of the storage LAN to those looking for more effective application performance.

Optical Fiber Channel storage LAN

If you use an optical fiber channel to store local networks, it is reasonable not only to describe the increasing cost of Optical Fiber Channel storage, conversion, and management, they also need to configure expensive host bus adapters for each server they connect to the storage LAN. Enterprises that use existing fiber channels to Store Local Area Networks will not encounter any obstacles. To achieve significant benefits of server virtualization, this complete fibre channel infrastructure includes vswitches and host bus adapters) must support NPIVN port ID virtualization) protocol. Currently, most products do not include NPIV.

Even with NPIV, VMware can transmit customer programs only between machines in one fiber channel region. This means that, although the user has achieved hardware independence on the server side, all physical servers in a group that can transmit client applications to each other are stored independently of an array or even a hard disk in a single fibre channel region ). Server hardware independence can cause dangerous multi-application hardware dependency in storage.

Optimize storage solutions for virtualized Environments

ISCSI Internet small computer system interface) or ip sanip storage LAN) provides the best storage solution in the virtualized server environment, which is not only the best in terms of obvious cost advantages, in addition, the availability, flexibility, and scalability of the virtual architecture are also the best. An iSCSI SAN storage system can also provide significant advantages for enterprises that use Wan disaster recovery virtualization. Snapshots can also be used at the storage level to copy data to a local or remote backup site.

In addition, The iSCSI storage LAN storage system has obvious Wan advantages over the fiber channel storage LAN. Fiber Channel storage Wan replication requires expensive FCIPIP-Fiber Channel gateway. You do not need to purchase, apply, operate, and manage additional systems for Wan replication for iSCSI storage LAN storage. ISCSI is a TCP/IP protocol that works locally throughout the WAN. Both Fiber Channel and iSCSI Wide Area Network replication can cause remote throughput reduction or packet loss. Wan or TCP/IP optimized devices used for iSCSI LAN storage can alleviate this problem. This type of Wan or TCP/IP optimization device has no effect on the FCIP gateway, or has little impact.

Trap 2: the dilemma of overprovisioning

Even with the correct storage LAN Solution, migrating applications to a virtualized environment sometimes slows down like crawling. If the server hardware configuration is correct, the administrator cannot tell why. In this case, storage is usually the cause of the problem.

Virtualization improves the efficiency of infrastructure by deliberately exceeding the configuration through management programs. Virtual customer applications are allocated with a physical resource that is lower than the best standards. This is based on the principle that it is statistically impossible for all applications to simultaneously demand resources. The principle of proportional use is generally feasible in practice. However, most storage LAN and storage LAN storage have used excessive configuration, and the result of dual excess configuration of physical storage resources is disastrous.

Because the storage infrastructure is really tight, conflicts have become a problem, and there are bottlenecks and cache overflow problems. What makes administrators feel more complicated is that these conflicts can occur on multiple layers of the storage infrastructure.

At the level of a single hard disk, the number of input/output request queues increases. This problem is more prominent When configuring a slow SATA hard disk. In SATA hard disks, the queue depth is generally 0 to 32 requests, while in SAS Serial Attached SCSI) or fiber channel hard disks, the queue depth is 256 to 512 requests. This means that enterprises that require application virtualization infrastructure need a LAN solution that does not limit their selection of backend hard disks.

At the storage LUN logical unit number) layer, the hypervisor itself generally Splits a physical storage pool or LUN into multiple virtual Luns. Then, these Luns are allocated to different virtual customer applications. These physical Luns cannot differentiate these customer applications. Excessive resource conflicts may reduce storage performance.

Similarly, hypervisor-level excess configurations can cause problems at the storage LAN infrastructure layer with host bus adapters, triggers, ports, and switches. These resources are often over-configured at or over-configured on the storage LAN. This combination of double overprovisioning not only reduces performance, but also causes request timeout and application crash.

Use virtual storage LAN storage to resolve excessive conflicts

One option is to disable the storage virtualization function in the hypervisor and manually allocate the LUN to each customer application. However, many vendors do not support this. In this way, key virtualization functions will be lost.

Another option is to solve this problem from the storage aspect to reduce the local excess configuration level in the storage LAN architecture. Using a physical storage LAN is complex and will significantly reduce the efficiency of the storage LAN for virtualized hosts. Using Virtual Storage LAN storage, this reconfiguration is not only simpler, but also can often treat management programs differently based on physical hosts to optimize the efficiency of the entire storage LAN.

Indeed, a virtualized storage LAN can also be used to distribute a single LUN to multiple storage resources, further alleviating resource conflicts. The Virtualized Storage LAN provides the performance of the storage LAN with the simplicity of network attached storage.

1. How to Build the safest server virtualization environment