SDN is disruptive because traditional networking suppliers like Cisco and Juniper, for example, have a more integrated approach where software and hardware are tightly tied together. SDN allows its software to run on “bare metal” boxes containing the world’s most used networking ASIC (Application Specific Integrated Circuit) from Broadcom.
Proponents of SDN cite ease of scalability, cost savings, and administrative simplicity. The SDN movement started as a “soft switch” and later became more widely known as the Open Networking Foundation supporting SDN and OpenFlow.
I had the pleasure of interviewing Mario Vecchio, Big Switch’s managing director for the APAC region about SDN and Big Switch. Ironically he started his career at Cisco. Excerpts below:
Why do we need SDN?
We need hyper-scalable networking as the number of users grows. For example, where a Telco ten years ago may have supported thousands of smartphone connections, now it's millions. Where Google or Facebook had millions of connections, now it's billions.
These companies know that they needed hyper-scalable networking and no matter how hard legacy providers tried, the reality was that the principles of OpenFlow and SDN were the right solution for them.
The other problem is that legacy systems may have been rock solid, but they were also inflexible. SDN can and does automate a lot of that complexity and has been proven effective by the world’s largest companies.
One of the biggest issues with legacy systems is that they need very high levels of expertise to keep them going. Operating systems in use today hark back to decade-old code where hyper-scalability was not an issue.
Google’s motivation and goals for its Project Juniper were simple – use commodity silicon (Broadcom ASIC), enable centralised control (SDN) and Clos fabric (multi-stage switching that connects switches everywhere as if they are one fabric). Oh, and they wanted to control their budgets, so automation was a key priority – they claim savings of 50-70% in hardware and operating costs.
How did Big Switch come about?
You have heard of the classic “outgrowing of academic environments” – well Big Switch came out of Stanford University in 2010. It was founded by the original team that invented SDN. Since then it has raised about US$100 million (completed C-round funding) to develop its open networking and has two products – Big Monitoring Fabric and Big Cloud Fabric. Its first commercial product was released in 2013. It is a private company.
Since then we have worked with most major companies — Microsoft, Verizon, Amazon, Intuit to name a few — and partners like Dell Technologies (our global distributor), A10, Red Hat, FireEye, BlueCoat, Palo Alto, VMWare, F5, Edgecore Networks, Broadcom, Canonical, Check Point, Citrix, Riverbed, ExtraHop etc. In Australia, we sell via the channel so it's Dell as a global distributor or local distributor Newgen.
Big Switch Networks aims to take the kind of hyper-scale networking found at Google and Facebook — one big switch — and bring it to enterprise and government data centres.
We have the five biggest banks as customers as they become completely digitally transformed.
I am pretty sure that there would not be a Tier 1 or 2 Telco carrier that is not commercially deploying SDN.
And government is a major user because of the compliance and visibility that SDN provides. It needs to get to a consolidated private network, and SDN enables this.
And what about you?
I am Cisco-trained, a 25-year veteran of networking and communications. I have been managing director from November 2015 and worked in senior roles for Siemens, Juniper, Dimension Data, Progility (UK Listed), and Communications Australia.
A lot of Big Switch staff have a similar grounding in networks and engineering. There is a lot of interest in SDN and this region already comprises more than 15% of our revenue.
Where is SDN and Big Switch heading?
Customers are driving us hard. As Marc Andreessen said in 2011, “Software is eating the world,” and we see software defining everything using bare metal, generic compute fabric. Efficiency is driving this and the need to be agile, scale quickly and be cost effective.
I think you will see Big Switch move from just the data centre to the software defined WLAN – then who knows. We cannot continue to use legacy technology developed in the 1990s.