Three advantages and best practices of data center network virtualization

The popular words in the IT industry often create the illusion that the network virtualization is just the IT industry for the virtual everything derived from the ancillary products, or just a fad. However, we can see that network virtualization has at least three points of benefit and there are several best practices that we should follow. So let's take a look at the significant advantages of network virtualization and its best practices.

The advantages of network virtualization

First, network virtualization can help make better use of network resources. Sometimes this means that the utilization of physical resources can be improved. For example, you can connect several virtual servers to a single physical network interface card (NIC). A single virtual server may not be able to take full advantage of Gigabit Ethernet (GbE) or 10,000 Gigabit Ethernet NICs, but resources will be better utilized if the NICs are shared by multiple virtual servers.

Network virtualization can also be used to split the traffic on overused resources. For example, IT staff can generate a virtual backbone as a logical framework for connecting virtual machines and the common hosts to which they are attached. This allows you to shunt the backbone network traffic from the physical switch. For example, suppose two applications are exchanging large quantities of data, such as databases and query systems. IT staff can transfer the two applications to the virtual machine, and the same virtual server as the host, and then set up a virtual network of two virtual machines on the server. This allows the two virtual machines to exchange data without having to go through any packets at all.

Of course, this concept also applies to network hardware. Sometimes, using virtual network hardware can free up the ports of the physical network connector.

The second advantage is that network virtualization is isolated. Sometimes, for security reasons, you might want to make certain routing protocols specific to a particular network zone. Many routing protocols or communication types can be isolated. For example, IT staff can isolate server backbone communications. Or, in another way, isolate HTTP traffic so that it avoids contact with other types of packets. The benefit of virtualization is that there is no need to bother deploying a dedicated physical schema to implement such isolation operations.

The third advantage is merging the ports. The case of using a single NIC for many virtual machines has been mentioned earlier. That is one of the consolidation patterns that can be used to improve utilization. However, consolidation can also be used to improve performance. For example, several NICs are bundled together as a logical NIC to provide more bandwidth than a single NIC can provide.

Best Practices for Network virtualization

Network virtualization adds a layer of abstraction and complexity to the network, so most network virtualization vendors offer a set of best practice guidance for the deployment program. Best practices should vary by product, but one of them is to be carefully followed by the IT staff regardless of product differences. That is, understand the performance of the network!

When creating an external virtual network (using a virtual network of physical network components), it is important to read the baseline data for network performance before and after the virtual network is erected. The performance of the virtual network is determined not only at the time of erection, but also on a regular basis.

Http://www.aliyun.com/zixun/aggregation/14216.html "> Performance monitoring is also important. Because when IT staff creates a logical virtual network architecture, it can sometimes affect seemingly unrelated network parts. For example, an IT person might find that the changes he has made make the internet overloaded. I have also seen that physical routers can cause packet loss in the event of network virtualization overload or saturation. Monitoring is the only reliable means. It ensures that network resources are used correctly under the condition of saturation.

Secondly, it is also to ensure that the network monitoring software is a virtual network can be recognized. Network monitoring software is usually located on a network server and uses the NIC of the server to send a ping and Simple Network Management Protocol requirements to the network device. Such software usually performs packet monitoring tasks as well. Unless it is a virtualized identification software, it is likely that it will not monitor parallel packet flows associated with virtual networks. Remember, one of the main reasons to create a virtual network is to isolate network traffic. However, if the reporting tool fails to respond to virtualization patterns due to design flaws, it may result in a distortion of the facts.

Even if the network monitoring software can observe the virtual network communication, only the virtual monitoring tools can reflect the global picture. For example, network monitoring software that is not dedicated to virtualized environments may not show the parts of the virtual network, or even if they are displayed, they cannot be identified as virtualized content. The actual results depend largely on the type of network virtualization and the monitoring software used.

The reason for this is that network parts like routers usually work only at three different levels: the hardware layer, the software layer, and the control framework layer. The level of implementation of the virtualization scenario can greatly affect the visibility of the virtual network to the management software. For example, if a software update allows a local router to support a virtual network, the router is likely to hide all the virtual network-related traffic from any management tool-not the tailor-made network monitoring tool.

Network virtualization can bring great benefits. Remember, however, that when you deploy a network virtual architecture, you follow recommended best practices to achieve the best results and avoid negative impact on network performance. (Source: http://www.searchdatacenter.com.cn/showcontent_49822.htm Author: Brien m. Posey translator: Sun Shao)