The network benefits of WAN Virtualization

Multi-path WAN reliability delivers better bandwidth economics, lower OpEx while enabling newer applications.

Having covered in recent columns what WAN Virtualization is and how it relates to WAN Optimization, let’s look at the benefits to the network that WAN Virtualization delivers for the WAN manager, which are substantial. [In an upcoming post, we’ll look at the benefits delivered by WAN Virtualization and the Next-generation Enterprise WAN (NEW) architecture beyond the WAN itself.]

WAN Virtualization drastically changes the price per bit of bandwidth for the enterprise Intranet, delivering anywhere from 30 to 100 times better cost/bit, by allowing enterprises for the first time to take advantage of Internet economics to augment and/or replace their expensive MPLS connections. While this game-changing, more than an order-of-magnitude improvement is perhaps the biggest benefit of WAN Virtualization, having covered this topic in some detail here and here, I won’t spend a lot of additional time on the economics now, except to point out that more bandwidth cures a lot of ills (not all, but a lot), and pretty much every enterprise buys bandwidth that fits their budget, rather than allows their budget to expand without limit to fit their bandwidth desires.

Beyond simply making bandwidth cheaper and offering enterprises a way to substantially reduce their monthly WAN bills, WAN Virtualization enables much more bandwidth per enterprise location, especially locations not connected via fiber links today, because of its ability to aggregate multiple, disparate links and use the combination as if it were a single connection. This additional bandwidth is usable for all applications pretty much all of the time, as opposed to the bandwidth savings delivered by WAN Optimization via data deduplication the second or successive time any given data is accessed from the location. WAN Virtualization offers the additional bandwidth even for initial transfers or for encrypted data flows. Some WAN Virtualization implementations allow you to use all of the bandwidth across multiple connections for a single application connection, or even a single TCP flow.

The bandwidth aggregation capabilities of WAN Virtualization enable enterprises to take advantage of the asymmetric bandwidth offered by inexpensive broadband connections. The asymmetry actually reflects the way that most traffic flows (from data center to branch), but a single ADSL connection alone, say, might not have sufficient upstream bandwidth to satisfy some corporate uses. By aggregating multiple connections, sufficient upstream bandwidth is available even for those enterprises that choose to eliminate a T1/E1 MPLS link at those sites.

This same aggregation capability is what allows for easy, cost-effective bandwidth scalability. Almost as importantly, bandwidth additions can now be incremental and done in days, rather than the weeks or months it can sometimes take to get an additional MPLS circuit deployed, or the months/years it can take to get fiber to some locations.

If revolutionary bandwidth economics is not the most important benefit delivered by WAN Virtualization, then the network reliability and application performance predictability it delivers is. As we saw here, the multiple network path reliability that WAN Virtualization brings is key to enabling the “Internet economics.” But that reliability has benefits beyond price/performance.

Using WAN Virtualization, the wide area network is now more fault tolerant even to errors you yourself might introduce (software upgrades, taking connections down for maintenance, etc.), thanks to the multi-path reliability combined with the fact that most WAN Virtualization implementations are overlays – software-defined networks, if you will, which sit on top of the routed network infrastructure rather than replacing or actively modifying them. Such a solution can enhance network reliability, and in particular application performance predictability, by eliminating single points of failure from even temporarily taking out the network and causing application connections to be broken.

With WAN Virtualization, the WAN manager needs to do far less troubleshooting of that portion of the network (the external WAN) where he or she has historically had both the least visibility and the least control. WAN Virtualization is not just a visibility tool to inform you of network problems or help track down the specific cause of a problem after it happened; it actually fixes the problems in real time before users even notice them. Of course, a good WAN Virtualization implementation will also give you rich visibility into WAN performance, to track down those problems which are not simply transient at your leisure, without impacting users, and to get money back from SLAs missed by your Service Provider. Lower troubleshooting costs translate into lower OpEx on the WAN, beyond the hard dollar cost savings.

With the best WAN Virtualization solutions, applications run “better” over the multi-network WAN fabric, with more predictable performance. VoIP, videoconferencing and VDI / Desktop virtualization over the WAN are examples of this, with benefits complementing and going beyond what existing WAN Optimization can do. In fact, the details of how each of these applications benefits from WAN Virtualization are worthy of a future column of their own! As an example, in our last column we touched on how WAN Virtualization can improve latency-sensitive VDI flows in ways that WAN Optimization technology cannot, via sub-second path switching in the face of network congestion and even replication of flows. For some interactive or real-time applications that are very sensitive to latency or packet loss, WAN Virtualization may be the difference between implementation success and failure, especially given the realities of today’s IT budget constraints.

WAN Virtualization, especially in conjunction with colocation, allows you to centralize the complexity in the network. As we know already from server consolidation, there are computing and security benefits to centralizing complexity in a small handful of locations. (We’ll actually cover the benefits that WAN Virtualization and the NEW architecture bring to computing and security next time.) But speaking just of the network itself, WAN Virtualization enables the centralization of network complexity and policy at a tiny number of data centers and colocation facilities, even for a large worldwide WAN with hundreds or thousands of locations, without sacrificing performance or security, and doing so in a low cost way that will scale.

The best example of this: scalable, compromise-free Internet backhaul. The NEW architecture and WAN Virtualization allow you to have all the security and simplicity benefits of network backhaul using a partial mesh hub-and-spokes design, with the higher performance and lower average latencies benefit of distributed Internet connectivity, in a manner which is less expensive and more scalable than either. Multiple low-cost broadband links at each branch can be used for scalability (and for efficient sharing with Intranet traffic) and hubs at the data center and carefully chosen colocation sites mean expensive, complex IPS and next-generation firewalls can be limited to only a few locations, maintaining security and allowing easier upgrades to and lower management costs for security solutions. For a partial picture of what this design would look like, see here. As I’ve noted a few times already in this post, this subject, too, is worthy of a whole column of its own, very soon.

A side benefit of WAN Virtualization is that it enables incremental, at-your-own-pace evolution of the WAN, rather than requiring periodic “forklift upgrades.” Older WAN connections need not be ripped out. If you desire, you can augment and keep older, expensive private WAN connections forever, or let them eventually wither away when they are no longer cost effective and sufficient total network reliability has been demonstrated. It’s now easy to take advantage of small amounts of expensive private bandwidth combined with larger amounts of cheap Internet bandwidth for the best of both worlds. Go ahead and keep spending, say, 60% of your WAN budget on the 10% of really expensive private (read: MPLS) bits. Spend the other 40% on the 90% of cheap Internet bits. Thanks to WAN Virtualization, you’ll not only have more bandwidth and lower cost per bit, but you’ll also have lower overall monthly WAN spend, and more reliability than putting all your eggs in the AT&T or Verizon or BT basket.

And while it’s not something that AT&T or Verizon want to hear, you now have a new source of funds for projects: the private WAN OpEx budget. In the same way that PCs, LANs, WANs and client-server computing spending grew much faster annually than IT budgets overall in the 1980s and 1990s by stealing from the mainframe/minicomputer budget, WAN Virtualization will enable enterprises to cap their expensive spending on MPLS and over time use that budget as a source of funds for not just adding bandwidth, but even other network priorities like enabling videoconferencing or migrating to cloud computing.

WAN Virtualization and the NEW architecture deliver such significant benefits to the network that even Enterprise WAN managers without an immediate need to cut costs or add bandwidth, or to find a solution to vexing problems like safely deploying videoconferencing, owe it to themselves to take a closer look.

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