I was out at Gartner Catalyst in London in September, speaking to IT professionals about their data center deployments. It was an enjoyable time engaging actively with other like-minded technical individuals that were interested in leveraging the boundaries of their technologies to drive greater business efficiencies and competitiveness.
The common theme across all the attendees I spoke to was the urge for containerization, flexibility of design and rapid deployment. These IT professionals were being tasked with reacting faster, and building more rapidly scalable environment. For their server and application needs, they all had turned to open solutions in Linux, leveraging operating systems such as Red Hat Enterprise Linux, Centos, Ubuntu, and orchestration tools such as Mesos and Docker Swarm to control Docker containers. The common point I saw was that all the compute infrastructure relied on open solutions that allowed for greater simplicity without sacrificing flexibility.
I would then ask these same IT professionals: “what do you use in for network infrastructure in these data centers?”
Universally, the response would come back: “Cisco” or “Arista” or “Juniper.”
I would push them: “Why?”
“Because it’s what we’ve always done.”
“It’s all we know.”
“No one ever got fired for buying Cisco.”
For that last statement, replace Cisco with Arista or Juniper and you get the gist.
For me personally, it’s difficult and confusing to hear these kind of statements. Entire companies are trusting their most valuable money-making and business-defining functionality to open solutions based on Linux (think Openstack and container solutions such as Docker, Mesos, and Kubernetes), but when it comes to the network that ensures service to these solutions, they keep picking closed, traditional solutions.
I look at traditional solutions vs. open solutions and I think to myself, every time someone picks a closed vendor, they have to ensure they have all the resources to manage an independent solution that doesn’t cleanly integrate with their operating system or applications. You need your own independent staff, independent monitoring tools, independent management tools and independent operational tools.
I then look to Cumulus Networks, and our plethora of open source solutions. I bring this up with all attendees and once they learn about our solutions, from Cumulus Linux as the open networking operating system, to NetQ as the operational workflow for modern data centers, and to Host Pack which unifies the network with server and application space, their eyes light up. They tell me they hadn’t considered these solutions beforehand, but now they wish that they hadn’t just gone with the status quo because it was what they always did.
They realized that, while going with these traditional vendors is certainly a popular option, it is far from the only option. And some went as far as realizing that they had done themselves a disservice by complacently picking their incumbent traditional network solution.
When all was said and done, I pitched my case for the new way and hoped that they would contact us when new opportunities arose, with the pledge that myself and my highly skilled colleagues at Cumulus could provide them unique insights into data center technologies and deployments that would keep their solutions simple, reliable and at their most competitive. And I earnestly hope that they come to that realization and defy the status quo. There are so many benefits to web-scale networking, such as increased scalability and better efficiency, that companies can take advantage of if only they would look outside of traditional options; the most popular solution is not always the best.
If you’re someone who’s looking to provide the best for their data center, ask yourself this question: “Why do I make the decisions that I do?” Is it because you’re following the herd and playing it safe? Or have you really considered which options would give you an unparalleled data center? My advice is that, if you have been picking solutions because it’s how you’ve always done it, you’re missing out on all the new innovations in the industry that you may not know are possible. I hope that this networking OpEd (or NetDevOpEd, if you will) helps to inspire people to think outside the vendor-locked box and give web-scale networking the chance it deserves.