Alcatel-Lucent Unveils SDN for the Masses
This week's announcement extends the vision to full automation and real-time tuning to optimize application performance.
It seems like a weekly event now to have some vendor issue a press release about something related to software-defined networks (SDNs). This week was Alcatel-Lucent's turn as they unveiled their SDN strategy, and I thought this announcement had some teeth to it.
ALU's SDN vision is an extension of its Application Fluent Network strategy that was rolled out in 2010 to be the company's network fabric vision. The original idea behind "Application Fluency" was to have the network be able to detect certain traffic types and then automatically configure itself accordingly. The low-hanging fruit for the company was real time applications such as VoIP and video, where they already have an embedded base of business.
This week's SDN announcement extends the vision of Application Fluency beyond recognizing different applications and basic configuration changes, to full automation and real-time tuning to optimize application performance. In addition to being able to do this for UC applications, ALU can now detect virtualized applications and apply a policy based on a pre-determined profile to enable the network to adjust to virtual traffic flows. The policies can be based on a wide variety of parameters including device type, users or different types of applications.
ALU says it will deliver the following enhancements in 2013:
* A set of RESTful APIs that improve programmability to support the OmniSwitch line
* Greater network recognition of applications through improved profiling and application analysis of virtual machine connectivity, to react faster to real-time network conditions
* Improved global control through the identification and automatic service activation of network infrastructure
In addition to the product enhancements, ALU's approach to SDNs is to address many of the limitations of the technology as it stands today. A recent study I conducted indicates that only 8% of organizations are not interested in SDNs, but the vast majority are still in early stages: A little over 80% are currently in the research or testing phase, meaning the industry as a whole is in the very early stages of this technology.
Despite all the hype, there are some "gotchas" that companies need to be aware of before heading down the SDN path. Specifically:
* SDNs can be complicated. One of my pet peeves is when "Google" gets mentioned regarding the future of networking. They've got literally hundreds of PhDs running their network, so if any company will be able to leverage SDNs today, it's them. Most companies, particularly small to mid-size ones, need high levels of automation to make sure when this stuff gets put into a running network, it works!
While the promise of SDNs is to simplify things, let's not kid ourselves--the path there many not be so simple. Think back to the early days of server virtualization. VM managers complained about VM sprawl, too many workloads to support, etc. Eventually the industry got a handle on this, but it took maybe half a decade for the promise of simplification to become a reality. Considering the importance of the network, I'm not sure the industry can afford a lot of early growing pains.
* Management difficulties. In some ways, this goes hand in hand with the above bullet, but network managers need tools that can manage not only physical infrastructure, but can span that virtual/physical boundary. There are some new tools that accomplish this, but ultimately there needs to be a tie-in to the existing tools.
* Co-existence versus rip and replace. SDNs are great, correct? Well maybe not so much if they require a new controller, new routers, new switches and new management tools. Ultimately, for organizations that aren't the biggest of the big, being able to deploy an SDN with existing network devices, and then migrating to it, is the best strategy.
These were some of the fundamental concepts behind ALU's SDN vision and related product roadmap. Automation of tasks was extremely important for the company to have right before they made the announcement.
The product enhancements will be released on the beefy OmniSwitch 10K and 6900 in the first half of 2013. In the second half of next year, ALU will roll out FibreChannel support for storage networks. The products are designed to be plug and play--or at least as plug and play as network gear can be--to enable customers to start small and then bring in more devices as needed.
The company also announced support for OpenFlow, but not until 2014. I've had a few people ask me about this and I don't really see an issue with this, considering how early we are in the SDN cycle. Also, contrary to some of the misinformation out there, one doesn't need OpenFlow to do SDN. It's certainly one way to do it, but it's not the only way.
Overall, I like the more practical approach ALU has taken to both network fabrics and now SDNs. The company, like everyone else, is targeting those big early adopters, but the ALU solution should have appeal to the masses as well.