What is WAN Virtualization?

Not your father’s link aggregation, it delivers reliability plus Internet economics for the Enterprise WAN

Current Job Listings

As you know if you’ve been reading this column for the last several weeks, I believe that we’re at the dawn of a NEW architecture that will revolutionize the enterprise WAN, enabling WAN managers to have much more bandwidth at much lower cost than their private MPLS WANs alone deliver today, with improved reliability and application performance predictability for traditional as well as newer applications, and safely and reliably enable the migration of more and more computing to the cloud – public or private – and do so in an incremental manner without forklift upgrades or breaking the bank on the WAN budget.

There are a number of technologies that are key to the Next-generation Enterprise WAN (NEW) architecture, and in this post we described what this combination looks like at a high level. The newest technology, WAN Virtualization, is by far the least understood, so let’s spend some time on it here.

WAN Virtualization enables network managers to use multiple WAN connections — existing private WANs such as MPLS, as well as any kind of Internet WAN links, be they DSL, cable, fiber, Metro Ethernet, etc. — to augment or replace individual private WAN connections. For some pictures of what the resulting network looks like, see here.

WAN Virtualization applies a RAID-like approach to solving the reliability and predictability issues associated with shared IP networks in general, and the public Internet in particular. It does real-time traffic engineering, reacting sub-second not just to link failures but also to network “brownouts” caused by congestion drastically.

By far the easiest way to understand what WAN Virtualization does, and at a high level how it works, is by analogy with what RAID has done for storage. RAID (Redundant Array of Inexpensive Disks) combines multiple cheap Seagate hard disks, wrapped with a layer of hardware and software intelligence, to deliver cheaper, higher capacity and higher reliability storage. Think of it as storage virtualization 1.0.

In a similar fashion, WAN Virtualization combines multiple WAN connections, including cheap Internet connections, wrapped with a layer of hardware and software intelligence, to deliver cheaper, higher-bandwidth capacity and more reliable WAN connectivity.

No one today builds monolithic, proprietary, single-disk storage solutions; they leverage commodity PC hard disk technology and use RAID to get greater capacity and performance, lower cost, and higher reliability. With WAN Virtualization, enterprise managers can take the same approach to corporate WAN economics that RAID gave them for storage economics, in both cases without sacrificing reliability. In fact, in both cases, it’s the reliability that the technology offers that is the key to making the economics work.

In the “sometimes a picture’s worth a thousand words” spirit, here’s a more graphical way to understand the analogy:

While there have been WAN link aggregation technologies for a long time now – MLPPP, for example, has been around forever - all of them had fundamental limitations on their usefulness prior to WAN Virtualization. They were either like MLPPP and required WAN links, which were uniform in type and bandwidth and had extremely low loss and jitter characteristics, or they would not deliver reliability in the face of network problems. Most other than MLPPP required that all flows use only a single WAN connection. If not, they would actually make application performance worse when there are congestion issues on any of the network connections or links. None have the ability to move flows off of a bad network path in the face of packet loss or excessive latency/jitter.

WAN Virtualization addresses all of those limitations of prior aggregation technologies, and then some. It performs per-packet traffic engineering in real time, and uses RAID-like techniques and other algorithms to deliver end-to-end reliability, predictable application performance for TCP-based applications, and ultra-reliable, cost effective, superior support for real-time application traffic like VoIP and videoconferencing.

It provides multipath multiplexing for both aggregating bandwidth and delivering reliability, dynamically engineering around network trouble - not just outright link failure, but high packet loss or excess latency - as it occurs. It adapts to such "network events" sub-second, a response more than fast enough to handle any application that can run on a cloud IP WAN like MPLS.

For real-time applications like VoIP and videoconferencing, WAN Virtualization can deliver ultra-reliable, cost-effective support, choosing network paths with the least packet loss and lowest jitter for such high-priority real-time traffic, and switching sub-second to a better path in the face of high loss or jitter. Where sufficient bandwidth is available, some WAN Virtualization solutions can provide still greater real-time flow reliability by replicating real-time traffic flows along a second path, suppressing duplicates at the receiving end, and so delivering “perfect” sound and voice quality even in the face of failures or congestion on one of the connections.

A properly implemented WAN Virtualization solution delivers as good — and usually better — reliability and application performance predictability as an enterprise can achieve with a single-vendor MPLS WAN.

WAN Virtualization requires the continuous gathering of rich real-time unidirectional information to deliver that end-to-end reliability and predictable application performance, which is why it can only be done as a two-ended solution.

Another two-ended technology, WAN Optimization, was built on the long-standing assumption that your existing network is as reliable as you need it to be, but bits are expensive – and the need, therefore, is to squeeze as much out of those thin pipes as you can. The insight behind WAN Virtualization is that there are lots of cheap bits out there, thanks to the Internet, they just need to be made reliable enough to be business quality.

Where WAN Optimization focuses on optimizing each application for the network, WAN Virtualization optimizes a more-affordable, higher-bandwidth network fabric for all applications.

WAN Optimization's data deduplication, specialized support for Microsoft's CIFS file access protocol, and other application-specific optimization techniques save network bandwidth but are best at providing application acceleration to enable centralized server environments like private clouds to provide acceptable application performance in the general case.

WAN Virtualization, thanks to its ability to aggregate bandwidth across disparate links, deals with the effects of packet loss, and quickly reacts to jitter and latency caused by network congestion, delivers better performance for first-time data transfers, improved quality and reliability for real-time apps like voice and videoconferencing, and offers more predictable performance for all applications even in the face of network “brownouts” which stymie other network reliability or application acceleration technologies.

Because the problems solved, and the techniques used, by WAN Virtualization and WAN Optimization are so different, the two technologies are highly complementary, and so they are both elements of the NEW architecture changing the game for the enterprise WAN.

A leading expert in WAN/LAN switching and routing, Andy founded Talari Networks, a pioneer in WAN Virtualization technology, and served as its first CEO. Andy is the author of an upcoming book on Next-generation Enterprise WANs.

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.
Now read: Getting grounded in IoT