Is network virtualization a reality or smoke?

Now you can take a sigh of relief. Software Defined Network (SDN) has become a reality, because Cisco has finally begun to "Embrace" it, which also marks the end of all kinds of "Cynicism. However, I cannot help but think of network virtualization because Cisco promotes the development of SDN. Network virtualization is different from SDN. It is completely different or slightly different. Different people will give you different answers. Network virtualization not only separates control, but also aims to separate the physical layer from the logical layer of the network. In short, any port on any physical device can provide a layer-2 network link for any logical physical device in your data center. This seems attractive, so I will ask this question: "Is network virtualization smoke ?"

Starting from the millennium bug

During this period, I was responsible for protecting 1500 Windows 95/98 systems from the threat of Millennium Bug. For this purpose, I seek help from Microsoft and hope to automate repair management by deploying its desktop management system SMS.

I encountered such a problem: At that time, SMS could not support its Novell Client/Server network staff. Microsoft promises to launch a product as soon as possible to solve this compatibility problem. After waiting for a few months, I still did not see Microsoft launching this product. This is my first time feeling "smoke". I have heard of shared items and free items, but not smoke. I soon learned that smoke is a product that has not been launched or has no shipment date.

Since then, I have no longer been excited about smog, nor have I changed its long-term architecture largely based on such software and hardware. However, the smoke of network Virtualization has shaken the author. Imagine that network Virtualization allows you to design a network completely isolated from the underlying hardware, the "pause, rewind, and record" functions provided by the Network for engineers will surely envy and hate experienced x86 virtualization administrators.

You may not be able to get the advantages of x86 virtualization integration, but this operation capability will surpass the x86 environment. However, the mature virtual network industry is actually a trap. We can imagine that this technology will face various technical obstacles. For example, where does the network management program of this technology run? How do physical and virtual networks interact with management programs and what new challenges will this model bring? X86 virtualization will not bring the same performance as physical servers unless the processor and chip fully support and enhance virtualization.

Even so, I don't have much hope.

That is to say, we will not see Cisco suddenly jump in and provide the same engineering guidance technology as the hardware optimization in the network device (as in x86 servers, although intel and other vendors are actively promoting related projects. But are these open switch-based methods powerful enough? Can 96-port Virtual Switches distributed on 20 physical devices (Far Away) provide the same performance as standard 96-Port 1 GB switches?

You must have considered other challenges. But the problem is that these are some tough issues that can be solved only by the joint efforts of the industry. VMware seems to want to take the lead in this regard through its network management program NSX. This technology has a huge potential, but so far it is still just smoke-this is my opinion. This concept is very exciting and some large suppliers are beginning to work hard. However, we still cannot build an architecture based on smoke.

What is your idea? Do you think virtual networks will only appear in sci-fi movies, or will this technology be implemented in advance?