The Software Defined Data Centre
Industry news sites and blogs have been active for quite some time with talk around solutions to support the changing environments of the modern data centre, and the challenges that virtualisation and cloud brings. With the steady rise of cloud based solutions, in both Service Provider and Enterprise environments, there is a need to support rapidly changing business requirements, increase mobility, performance, security, scalability and availability, oh and if you can reduce operational costs as well, that would be great!….. Not a simple task.
As it stands right now, traditional networking solutions are not cutting it, which is the reason why lots of new solutions; companies; and protocols, have materialised over the past few years to try to solve these challenges. At present it is still wide open with no clear winner.
At present, these solutions can be broken down into two categories:
1) Host based solutions – A software component is added to each host / virtual machine, allowing it to communicate with all other “enabled” hosts regardless of the underlying infrastructure (smart edge).
2) Network based solutions – The physical network is manipulated in some way to enable communication between hosts / virtual machines, without them being involved in the process (smart core).
In both scenarios there are multiple options available, including proposals from companies like Nicira and Big Switch, (draft) standards like VXLAN, NVGRE and TRILL, as well as other vendor specific solutions like FabricPath (Cisco), OTV (Cisco) and QFabric (Juniper). An element that is common across multiple options is Openflow.
For those that do not know, Openflow is an open standard protocol that decouples the control and forwarding planes of a network so that traffic flows are controlled from a central point and pushed down directly into a devices forwarding table. This is obviously different from a traditional network where the control and forwarding planes are present on every single device. (Please see http://www.openflow.org/ for more info).
Simply put, it allows for the easy deployment of innovative routing and switching solutions that can support multi tenancy, virtual machine mobility, and increased security and not be dependent on the underlying hardware.
From reading a few blog posts and whitepapers, the questions I ask myself are:
a) Does Openflow have real mileage outside of academic institutions and large cloud / service providers?
b) Should Vendors with an investment in proprietary hardware / software be concerned?
Does Openflow have real mileage outside of academic institutions and large cloud / service providers?
At present Openflow seems to be very well suited to the groups mentioned above, for two main reasons:
1) Openflow abstracts the network intelligence to a centralised function and as such provides more flexibility and choice of the underlying hardware. This could have potential financial benefits, as there is a greater choice of hardware available without having to worry about feature parity.
2) Because Openflow is not restricted to the underlying hardware, there are several technical enhancements, including virtualised workload mobility, multi-tenant security, and increased scalability.
Openflow has an appeal within academic institutions, as it is a flexible way to test out new features and configurations, which may not be available on a traditional vendors’ infrastructure. The Openflow website description lends itself to this environment as well “OpenFlow is an open standard that enables researchers to run experimental protocols in the campus networks we use every day”
The issue I think the mass market has with Openflow is exactly the opposite of why Service Providers and Universities like it. Most customers want to buy a wrapped solution that is proven, tested, easy to implement and manage, and has the backing of a vendor or systems integrator - allowing these customers to focus on their business priorities, rather than an internal science project. You could argue that these customers should consider a cloud service constructed with Openflow, thereby benefitting indirectly, but that is a different topic entirely.
Should the big hardware and software vendors be concerned?
I do not think that the traditional vendors should be concerned, rather they should embrace the imminent change and lead the pack, as a there is a potentially valuable market in waiting.
Quite a few of the main hardware and software vendors are involved (as members) with the Open Network Foundation (ONF) and have announced support for Openflow within their product lines, these include Juniper, Extreme, and HP.
Until recently, Cisco has remained very quiet on the subject, as naturally they have the most to lose if they do not execute their strategy correctly. They have now started to speak up on the subject with John Chambers recently saying that Cisco has “re-invented itself” and that they are a believer in “software defined networking”.
There is also a rumour going around that Cisco is once again incubating / investing in a start-up (called Insiemi) that has a focus on “Software Defined Networking”. This rumour is even more interesting due to the familiar names who are involved, which include Mario Mazzola; Prem Jain; and Luca Cafiero. All three of these senior industry figures were involved in previous Cisco spin-in success stories, like Andiamo Systems, which produced to the MDS line of Fibre Channel switches, and Nuova systems that created the Nexus product line and kick started Cisco’s Unified Computing System (UCS). What Insiemi is developing and how it will integrate into the existing Cisco product portfolio will be interesting and challenging, but based on how successful the MDS, Nexus and UCS products have been they are probably odds on favourites to succeed.
Summary:
I think Software Defined Networking solutions will be ready for the mass market at some stage, and will yield benefits, but it will require products, solutions, and leadership from the normal suspects (Cisco; Juniper; VMware; etc.). It is going to be an interesting couple of years, with lots of options available, and a lot more competition than we have seen in the past – all of which can only be a good thing for customers.
This is a guest blog post. Views expressed in this post are original thoughts posted by Nathan Deane, Pre-Sales Consultant at Imtech ICT Ltd. These views are his own and in no way do they represent the views of the company he works for.
