2011 Network Virtualization Predictions
“Computers in the future will weigh no more than 1.5 tons.”
– Popular Mechanics, forecasting the advance of science, 1949
As a wit once said, forecasting is hard – especially with respect to the future. But sometimes the future is teed up so clearly you just can’t help but get it right. For networking, this is one of those times.
Here’s why: 2010 marked two momentous occasions in networking:
- Network virtual machines came to market. In 2009 there were hardly any; now there are network VMs from Vyatta, Riverbed, F5, Checkpoint, Citrix, Fortinet… if you can name a network device company they’ve probably recently released their product as a VM.
- Intel reached 10Gig. With the release of the Nehalem platform in January, throughput on a single-socket x86 box surpassed 10 Gb/s — exceeding the requirements of the vast majority of the Layer 3-7 market.
Given this factual background, 2011 is clearly shaping up to transform the networking industry. Here are a few predictions:
Network Edge: Device Consolidation
2011 will bring the beginning of the end for single-purpose devices in the branch office and campus as virtualization enables these functions to consolidate on a single device. With all the high-visibility activity around cloud and data centers lately it’s easy to forget that Edge Networking is an $8B industry. 2011 brings a windfall opportunity for organizations to significantly reduce their edge network bill by deploying network virtual machines instead of single-purpose devices.
Think about the case of large distributed enterprise. Each branch or campus has a router, a VPN/Firewall, a WAN Optimization device, perhaps an IPS or Web Filtering appliance. For a 500-office company that’s ~2,000 – 3,000 discrete network hardware appliances. That number can be cut down by 80% by using network virtual machines on a single device, saving the enterprise untold millions of dollars in otherwise redundant hardware installations.
It’s already happening. For example, Riverbed is now shipping a Vyatta VM on their appliances. What was a single-purpose WAN Optimization device is now a WAN Opt / Router / Firewall / VPN / IPS / URL Filtering solution. And there are a lot more examples waiting in the wings.
Cloud Datacenter: Layer 3-7 Steals The Spotlight
In 2010, Layer 2 earned “shiny object of the year award” as the vSwitch came to market. It was necessary; virtualized servers require L2 functionality directly underneath the VMs in order for traffic to be managed between the server and the switch controller. But what about the bevy of other network services that modern IT relies upon, such as routing, encryption, tunnels, firewalls, traffic management, etc? These are up in Layers 3-7, and the need for them doesn’t go away with cloud architectures.
2011 is the year virtual L3-7 solutions get rolled out in cloud datacenters. As noted above, the VMs are already in market; network architects have had close to a year to work them into their plans. And the first real operational uses are already up and running.
The business agility that virtualized L3-7 offers is tremendous. No more specialized boxes. No more ladders, cables, and underutilized CapEx. Just push a button and Bing! there’s the network service you need, courtesy of a VM running on part of an x86 server that’s already racked.
Big Iron Networking: Intel Eyes 100Gb/s
Speaking of x86, the next releases from Intel (from Sandy Bridge onward) will continue to demonstrate how much focus they’ve put on networking performance. It would be completely inappropriate to say Intel’s “chipping” away at the problem — not only would that be a bad pun, it fails to capture the magnitude of the attack that Intel has launched on the networking space. To wit: Packet-processing performance on x86 has increased 100X (not 100%, 100X) in just the past four years.
With faster systems, more cores and software architectures that uniquely take advantage of the new technologies, it’s clear that Intel is doing to the networking market what they did to the Unix server market — which is to say they trounced it mercilessly. And they have only just begun for networking. The relentless competitive drive of the world’s largest semiconductor company just won’t stop.
That’s the 1-2-3 of Vyatta’s predictions for 2011. There are others we could make, but it’s hard to conceive how they could rock the industry harder than these first three will.
Entry filed under: Uncategorized.