The post that started it all

Recently I read a post that reflected exactly my line of thought. I was about to comment that post when I decided to write it here.

The post was OpenFlow BIOS does not server make, it dates from last April so it's not a new post but its getting a habit for our previous and current work to be compared to OpenFlow.

OpenFlow is to us what a BIOS is to a virtual machine. It acts as an underlying abstraction to the hardware by making it look like a simple forwarding engine. This is desirable but not a requirement. In fact, many OpenFlow 1.0 implementations are limited and gives us the same capabalities (VLAN management, translations, etc.) we had back in the original UCLP days back in 2001. I think this is the reason why these solutions are being compared even now. However, in our solutions we now support OpenFlow network elements and they don't bring much more functionalities to the enduser but it does make the job alot easier for us as we don't have to write special drivers to modify ACLs, Routing Policies, VLAN management etc. for every device from every vendor.

In this line of thought I invite you to read:

• You don't need OpenFlow to Solver Every Age Old Problem.

Moreover, discussions on OpenFlow often brings the topics on network virtualization which basically makes me uncomfortable. An example would be statements like

"Network virtualization with OpenFlow will be the killer app"

We have been working on network element virtualization for a couple of years and there are many approaches to do so, from isolated permission management and APIs down to full virtualized network OS stacks which as valid approaches based on the user's needs for isolation and manageability (think OpenVZ vs Xen/KVM). For us and what we have been doing in the last years, OpenFlow only represents a minor improvement and some better low-level access to the hardware (which we appreciate).

• The Verdict: Either OpenFlow Matters a Lot, or Not at All

I must say I find myself on Arista Network's side in saying that OpenFlow is currently overhyped and while it does bring many benefits to all of us in the network virtualization space t it is no silver bullet towards network virtualization and infrastructure services.

About Inocybe Technologies

It is always a hard thing to do the first post when you start a new blog. The old one was deleted when Blogger moved to Google a while ago, and since most of the posts were obsolete I decide to reopen a new blog for Inocybe Technologies inc. I have been sharing ideas and doing a variety of presentations for the last 5 years but never seemed to find time to post these ideas on a blog. I recently decided to change this behavior and let people know and get additional comments.

Quick Introduction

My name is Mathieu Lemay, and I'm the founder and currently CEO of Inocybe Technologies. In the past 10 years I've specialized in software-defined networks and cloud computing. In 2001, I worked at the Communications Research Centre where we participated in a CANARIE program called User Controlled LightPaths (UCLP). This was an opensource software that was meant to allow user empowerment of network configuration. In 2005, the concept was pushed even further with the "Lego" approach to networking in UCLPv2. One could get network pieces and try to orchestrate them into solutions for end-users. This concept was pushing network operation and service composition, In 2007-2009 we tried to commercialize UCLP under the Argia brand-name. Sadly, this turned out to be a failure because the solution had been designed from research and didn't meet market expectations. In the last two years we did a complete rewrite and worked with industry players to come up with a Platform for Virtual Networking which can be accessed on Inocybe's website.

Purpose of this Blog

While I will present some of the products and solutions from Inocybe Technologies inc. in the posts, the real purpose of the blog is to share my views and opinions on the different solutions and technologies. Cloud computing and Software defined networking are current trends and I will talk about a variety of aspects found in these types of systems ranging from programming languages we use like Scala and Erlang, Infrastructure as a Services concepts, OpenFlow and network protocols. These posts will share my experience in working with these technologies and solutions.

Please comment my posts

People are free to comment my posts and I welcome constructive criticism. However this is a corporate blog and I expect the comments to be professional and not simply technology rants. I will monitor this blog regularly and try to participate in these discussions as much as I can.