Archived Content

The following content is from an older version of this website, and may not display correctly.

It was a holiday-shortened week in the United States, but there was still plenty to talk about in software-defined networking (SDN) and network functions virtualization (NFV).

1. NFV Part 1: SK + NSN = EPC

SK Telecom demonstrated a proof-of-concept evolved packet core (EPC) for mobile networks. The vendor SK chose to work with was Nokia Solutions and Networks (NSN; formerly Nokia Siemens Networks), which SK said was the only vendor in its trials to successfully demonstrate a complete and virtualized EPC in the cloud.

2. NFV Part 2: CloudNFV Strengthens Carrier Ties

Separately, CloudNFV inched closer to having real products. This is the effort led by analyst Tom Nolle of CIMI Corp., as you'll recall, and its main objective is to create a cloud-based NFV architecture that pays particular attention to the management plane, which Nolle was afraid would be neglected as vendors fleshed out their NFV ideas.

CloudNFV is now in talks with two Tier 1 carriers interested in using the architecture, Nolle says. And as he tells Light Reading, CloudNFV's ongoing work with carriers is leading the group in some new directions. CloudNFV is now developing four components it hadn't considered before getting carrier input: a node builder, a service builder, a deployment manager, and a service operations manager.

A virtual EPC is on the CloudNFV radar, too. CloudNFV already handles voice traffic, using the cloud-based IMS technology developed by Metaswitch's Project Clearwater, and the CloudNFV group is working on deep packet inspection (DPI)-based traffic management. Both these pieces are steps toward developing that virtual EPC.

3. ScaleXtreme Reaches Out to the Clouds

Startup ScaleXtreme released cloud-management software for public clouds. Advanced Cloud Management, as the software is cleverly named, aims to automate the deployment of applications across multiple clouds, so that network administrators don't have to care (as much) about where a particular job is running.

It's primarily used in greenfield cases, where a customer has just started using public clouds and doesn't have a deeply ensconced interest in a particular management framework. Plenty of big enterprises are in that situation, says Nand Mulchandani, ScaleXtreme's CEO

But in general, ScaleXtreme's pitch is that big-company management software isn't tailored for the public cloud. "An Amazon instance costs you 9 cents an hour to run. Why would you buy BMC BladeLogic for it, when it costs $450 a server to run?" Mulchandani says. By contrast, a more expensive deployment of Advanced Cloud Management goes for $15 per server per month.

Key to ScaleXtreme's software is the use of APIs to tap the capabilities of public clouds and to see what's going on in there. That "cloud whisperer" ability has value. Lyatiss, for example, offers a way to see the virtual network you've built in Amazon Web Services and recently added analytics on top of that.

4. The New Extreme

The acquisition of Enterasys boosts Extreme Networks' switch market share to 2.4 percent, according to Dell'Oro Group — but that's good for a whopping fourth place in market-share rankings. As Network World reports, Extreme gains in areas other than raw tonnage. Enterasys' wireless-LAN joint venture with Siemens Enterprise Communications (now called Unify) will replace a less-than-ideal OEM arrangement with Motorola Solutions.

And Enterasys brings a network management system, NetSight, that Extreme plans to adopt. Users of Extreme's Ridgeline management software will be migrated to NetSight eventually.

The fate of Enterasys' switch technology is still undecided. Enterasys' designs include an ASIC, but that could be abandoned if the same capabilities can be wrung out of merchant chips.

5. SDN and Big Data

From Mike Bushong at Plexxi comes this analysis of big data's potential role in SDN. Ideally, SDN creates a feedback loop where data about network performance triggers automated responses in the network — moving a workload to a less congested area, for instance. But that brings up question such as how much data to use, or how to use it.