OpenStack Summit 2016: Key takeaways & Rocket Turtles

A few weeks ago, several of Cumulus Networks’ passionate employees headed to Barcelona to connect with other cloud-enthusiasts at the 2016 OpenStack Summit in Barcelona. We were there to visit our customers, partners, talk about our solutions, and connect with other industry experts. Here are the highlights.

If you were lucky enough to attend the event, you may have met one of our infectious leaders (CTO and Co-founder) JR Rivers. He was manning our booth along with the EMEA team and a few marketing and product members. The team was there to connect with partners, talk about our technology, and of course, pass out our new rocket turtle.

We’d like to give a special shoutout to JR who unwrapped more than 800 #RocketTurtles! They were an instant hit, and have been spotted in various parts of Barcelona.

By visiting our booth or our partners’, Mellanox, Red Hat, Dell and Midokura, attendees were able to enter a chance to win a Lego Mindstorm EV3. Congrats to our winner, Jules Chevalier!

One of the highlights of the event for us was connecting with other industry folks in Continue reading

Cumulus Linux Network Command Line Utility

Here at Cumulus Networks, we believe that network engineers are the real heart of an organization. They’re the ones managing switches, running the data center, and generally keeping an organization moving efficiently and securely.

We also believe web-scale networking and the associated benefits should be accessible to everyone and the best way to make that happen is to leverage the power of disaggregation and native Linux. Although web-scale networking is very flexible, agile and offers many benefits, there can be a learning curve as Linux uses separate, independently developed applications which each have their own syntax to configure the switch.

So how do we bridge these two beliefs? Allow us to introduce Cumulus Linux Network Command Line Utility (NCLU). Scheduled for our early December 3.2 release, NCLU empowers and quickens the learning curve so all network engineers can benefit from web-scale networking while integrating with and still supporting the traditional Linux methods. In short, NCLU makes Cumulus Linux easily accessible to everyone.

What is Cumulus Linux Network Command Line Utility?

NCLU is a command line utility for Cumulus Linux that rides in the Linux user space as seen below. It provides consistent access to networking commands directly via bash, Continue reading

Introducing the Solutions Marketplace

Traditional enterprise networking is under siege — threatened by choice, by open source, and by open standards. The same revolution that made Linux the standard for server operating systems is now happening to network switching. With over 1.5 million ports in production, 50+ certified hardware platforms across 8 hardware vendors, Cumulus Linux® is the de-facto platform for Open Networking, and a perfect example of what the next generation data centers should include.

But the age-old claim that “It’s Just Linux, you can do whatever you want!” can complicate solving specific problems customers have in the enterprise. Based on feedback from community members, we’ve created the Solutions Marketplace on the Cumulus Networks Community Website. The Solutions Marketplace is a repository of community-submitted projects, user space applications, automation scripts, and extensions to Cumulus Linux. This enables collaboration and fosters innovation through a common platform to develop upon openly and freely using Cumulus VX.

The Solutions Marketplace with Cumulus Linux expedites the path to production due to the availability of existing community expertise. Best practices are shared, which means you don’t have to start from zero when building out your data center. A disaggregated hardware/software model enables flexible environments and leverages Continue reading

25 is the New 10

I am continually amazed at the productivity of the Cumulus Networks development team who recently collaborated with the Mellanox development team to do some amazing things in this new Cumulus Linux 3.1 release. Besides innovating on a number of important new software features, they added support for five new switches from Mellanox, including the first native 25 Gigabit Open Ethernet switch as well as the highest capacity 10/100GbE switch on the market.

The Mellanox SN2410 is the industry’s first generally available switch with native 25 Gigabit Ethernet (25GbE) ports. Working for *the* provider of 99% of all 25GbE NICs sold worldwide, I can say with confidence that, until now, all 25GbE servers have been connected to 100GbE switch ports via breakout cables. The SN2410 changes all that by providing 48 SFP28 ports that can natively operate in 1G, 10G, or 25G modes, which is great for cutting-edge deployments while providing backward compatibility for legacy devices. Just like 10GbE SFP+ ports, the 25GbE SP28 ports can utilize inexpensive passive copper direct-attach cables.

Who needs 25 Gigabit Ethernet?

Without my naming names, you can safely assume that the hyperscalers and other early consumers of 10GbE are now moving to 25GbE Continue reading

Riding the Wave of Linux Greatness

We at Cumulus Networks firmly believe that Linux networking is awesome, as it provides a high function, extensible platform for networking. Linux has always been the platform of choice for decades; every system today starts with Linux as its base and builds around it.

With the advent of the virtual machine and container networking, the constructs that used to be relegated to physical switches are applicable on servers, we thrive on the fact that Cumulus Linux networking IS Linux networking.

So, what does it take to bring that the greatest and latest in Linux to you? It takes two things really:

• The latest Linux kernel, with all the new features and facilities.
• The latest system libraries that enable those features for applications and the latest versions of the huge ecosystem of Linux software packages.

Kernel at the Core

The kernel is the center of the Linux operating system. We work closely with the Linux kernel community to add new networking features or extend Linux networking APIs for NOSes. With Cumulus Linux 3.0, we started with Linux kernel version 4.1 and networking patches from even more recent kernels. This provides networking applications with the latest Linux APIs, Continue reading

Making Networking Great Again: Leveraging ifupdown2 in the Data Center

I love election season, mainly for all the great slogans. Every candidate is trying to find a way to catch the attention of the electorate in order to get their ideas across. If people don’t know the benefits of a new solution, they’ll be hard pressed to understand how much better life can be.

The same can be said for Linux networking when ifupdown2 came along. This article describes the improvements made to ifupdown2, but it doesn’t describe the excruciating pain of having to run the classic ifupdown. I feel obliged to join this campaign cycle to wholeheartedly endorse ifupdown2 and tell you about how it’s making networking great again.

I was recently simulating a data center environment with Vagrant to test scalable architectures. I was trying to leverage ECMP via the new Routing on the Host feature on an Ubuntu 14.04LTS server over a Cumulus Linux spine/leaf Clos network. One requirement for this feature to work is peering BGP between the Ubuntu server and the first-hop leaf. Sounds simple, right? I had already peered BGP throughout my entire Cumulus Linux switch network, and since Ubuntu is also a Debian-based distribution, it should have been a trivial task.

Read Continue reading

Democratizing Capacity (or How to Interpret Cisco math)

Mid-December, Cisco held its Financial Analysts Conference. Shortly thereafter, I was fielding questions that summed up to “What is this all about? Cisco is cheaper than bare-metal? Really?”; check out slide 4 in this presentation.

Let me begin by clarifying that Cisco is not more affordable. Period. We’ll get to that analysis a bit later… the bigger point is that bare-metal networking is more than being affordable; it’s about giving customers degrees of freedom, transparency, and choice that they deserve in a mature industry.

In the dark ages of computing (aka 1983), a customer running IBM DB2 had to buy an IBM mainframe (complete with cables, disks, power distribution, memory, and IO) to go with the application. The compute industry has matured to a point where DB2 runs on hardware ranging from mainframes through p-series down into non-IBM x86 platforms hosted on operating systems including z/OS, Unix, Linux, and Windows. Application independent from OS independent from hardware; degrees of freedom and choice.

Circling back to networking, a significant number of products today, including Cisco’s Nexus 3000 and Nexus 9000 platforms, are based on industry prevalent networking silicon from companies like Broadcom. Optics and cables branded by networking incumbents are Continue reading

Independence from L2 Data Centers

We’ve all been there. That “non-disruptive” maintenance window that should “only be a blip”. You sit down at the terminal at 10pm expecting that adding a new server to the MLAG domain or upgrading a single switch will be a simple process, only to lose the rack of dual-attached servers and spend the rest of your Thursday night frantically trying to bring the cluster back online.

If I never spend another evening troubleshooting an outage caused by MLAG, I’ll die happy!

While MLAG provides higher availability than single attaching or a creating multi-port bond to a single switch, it comes with the cost of a delicate balancing act. What if there was a way to provide redundancy without MLAG’s fragility and its risk to maintenance windows?

We at Cumulus Networks have seen many of our customers solve these problems by leveraging Cumulus Quagga, our enhanced version of the routing suite, on their server hosts, so we’ve decided to call it Routing on the Host and make it broadly available for download.

By leveraging the routing protocols OSPF or BGP all the way to the server, we can resolve that MLAG problem once and for all.

Over the last five years, Continue reading

Is It Real, or Is It Cumulus VX?

Phew! Cumulus Linux 3.0 has just been released! A big shout out to all of my engineering colleagues who worked so hard to make this happen. JR Rivers gave an overview of all the goodies included in 3.0 in his recent blog post. Stay tuned for more blog posts from other engineers for details on all of those new features.

But Cumulus Linux isn’t the only beneficiary of all the 3.0 work. Cumulus VX, our free virtual machine-based version of Cumulus Linux, also has some pretty cool new tricks. When we launched Cumulus VX last August we thought it would be a way for people to get hands on with a Linux-based switch operating system, in their own environment and without any commitment. Boy, Were we surprised at how it quickly became so much more. With over 3,800 unique users, Cumulus VX is being deployed in all sorts of ways we never dreamed of. As just one example, existing customers are using it to validate their configurations before upgrading their physical switches from one release to another.

What’s New?

And that brings me to the first change we’ve made: concurrent releases. Our plan from now on Continue reading

Silicon Choices

Cumulus Networks has always strived to provide our customers with choice. And now, Cumulus Linux 3.0 has been refactored to make the user experience fun and easy … and at the same time bring you, even more choice:

• In hardware platforms, from a variety of manufacturers.
• In CPU architectures — x86 and ARM.
• In Broadcom networking chips for a variety of use cases — 1G, 10G, 10GBase-T, 40G and 100G switches with Helix4, Hurricane2, Trident-II, Trident-II+ or Tomahawk chips inside.
• And now, for the first time, choice in silicon vendors with the introduction of Cumulus Linux-powered 40G and 100G switches with Mellanox Spectrum chips inside.

Shrijeet Mukherjee (our fearless engineering leader) kicked off the Cumulus Linux 3.0 development cycle with this as our main goal — to offer even more choice to our customers. And then this happened…

So we took on the challenge of unifying the user experience across this sweeping range of hardware platforms and switch silicon without muting the unique prowess and feature richness of any of them.

How did we do this? By deploying the three dragons in our arsenal: ONIE, switchd and the Linux kernel itself (which is now better fed by Continue reading

Hey Rocket Turtle, Pimp My Ride!

This colloquialism for “make my vehicle better” is an appropriate perspective on our recently released Cumulus Linux 3.0, or as we like to say around the office, “3.0.” Our engineering team looked at the upcoming market changes and decided to give Cumulus Linux a pretty sweet makeover.

Starting with the “IP mindset” that prevails in modern deployments, our team worked with the Linux kernel community to add Virtual Routing and Forwarding (VRF) to the kernel and hardware support to Cumulus Linux. VRF is coupled with BGP unnumbered interfaces as an even simpler way to deploy multi-tenant dual-stack IPv4/IPv6 networks.

In parallel, we looked at the image installation and upgrade mechanisms, revamping the build, packaging, and base installer. As a result, 3.0 is based on Debian 8 (Jessie) and Linux kernel 4.1 tied together with an entire system that enables the development flexibility of Linux coupled with the testing and support required for wide-scale, enterprise production deployments.

All of this functional horsepower is applied to seven new hardware platforms continuing Cumulus Networks’ industry leading support for Open Networking systems. These platforms cover the gamut of speeds, feeds, and functions; introducing Mellanox Spectrum alongside Broadcom Tomahawk and Continue reading

MLAG – An Implementation for Everyone!

I typically don’t to get up on a soapbox and preach the awesomeness of Linux networking, but I think I’m going to make an exception for this one topic: MLAG.

Yes, MLAG, that wonderful non-standard Multi-chassis Link Aggregation protocol that enables layer 2 multipathing from the host to gain either additional bandwidth or link resiliency. Every vendor that supports MLAG does so by using their own custom rolled implementation of it, which means Vendor A’s version of MLAG cannot interoperate with Vendor B’s version of MLAG. So I can’t have one switch be an “X” box and another be a “Y” box and expect the two to be part of the same MLAG configuration with a Dell server.

That ends today (arguably I could have said, that ended January 2015 when Cumulus Networks shipped with MLAG support in Cumulus Linux 2.5, but I’ll get to that in a bit).  Several weeks ago I was with my colleagues Shrijeet Mukherjee and Tuyen Quoc giving a talk about how “Linux Networking Is Awesome” at the 2016 OCP Summit. During our standing room only talk, we explained how Linux networking has become the de-facto networking stack in the data center (and Continue reading

Microservices Network Architecture 101

A new god is rising in the world of application development – Microservices

The new god promises if not happiness in the next life, scalability, agility and fault tolerance in this life. At the heart of all this, is a simple, age-old axiom that is a key design goal of Unix: do one thing, and do it well. In the evolution of application architectures, single monolithic applications made way for client-server applications, which in turn made the way for microservices. The upending of the old world continues in data centers.

Communication is at the heart of this new religion (one popular theory of the etymology of the word religion is the word “religio” which means “to reconnect”). Every religion and every new technology introduces its own new vocabulary.

Microservices are no different!

In the domain of communications, the new lingo involves things such as MacVlan, IPVlan, Weave, Flannel and Swarm, to just name a few. What are they ? How are they connected ? Is IPVlan a new encapsulation format ? If it’s not a new encapsulation format, what is it ? If it is a new encapsulation format, how is it related to VxLAN ? Why were they invented ? Which one should I use ? What Continue reading

Microservices Network Architecture 101

A new god is rising in the world of application development – Microservices

The new god promises if not happiness in the next life, scalability, agility and fault tolerance in this life. At the heart of all this, is a simple, age-old axiom that is a key design goal of Unix: do one thing, and do it well. In the evolution of application architectures, single monolithic applications made way for client-server applications, which in turn made the way for microservices. The upending of the old world continues in data centers.

Communication is at the heart of this new religion (one popular theory of the etymology of the word religion is the word “religio” which means “to reconnect”). Every religion and every new technology introduces its own new vocabulary.

Microservices are no different!

In the domain of communications, the new lingo involves things such as MacVlan, IPVlan, Weave, Flannel and Swarm, to just name a few. What are they ? How are they connected ? Is IPVlan a new encapsulation format ? If it’s not a new encapsulation format, what is it ? If it is a new encapsulation format, how is it related to VxLAN ? Why were they invented ? Which one should I use ? What Continue reading

With the Hype around SDN- Is it Worth it in a Data Center?

With all the hype about SDN, we wonder: Is it worth deploying it in a data center? What goals does SDN meet? Is there a simpler way to achieve those same goals?

We think there is a better way, and it’s called Open Networking. And Open Networking is available today.

Vivek Venkataraman (my colleague and co-author of this blog) and I recently presented “Achieving SDN Goals through Open Networking” at the Open Networking Summit in Santa Clara, CA. The talk was well attended and interactive. We examined the goals of SDN:

• Controlling costs
• Achieving agility and flexibility
• Easily enabling and encouraging innovation
• Using the network as a platform

You can achieve all of these goals today in a more pragmatic and efficient fashion using Linux and Open Networking.

Controlling Costs

Reducing both CapEx and OpEx is always an important goal for any enterprise. First, Open Networking reduces CapEx by encouraging vendor competition and giving customers choice, all the way from the optics and silicon up to the OS and applications. The modular choice allows data centers to be designed to their exact business requirements, so customers pay for and deploy only the necessary hardware, features and applications that Continue reading

Will OpenStack Stay Disreputable?

“Keeping It Dirty”

I’ve lived in Durham, North Carolina since 1999 — I love it here, and I’ve finally found home. It’s been recognized as Tastiest Town, a Different Kind of Silicon Valley and one of the Best Places to Live. But it wasn’t always like that. Durham rose up from the ashes of failed tobacco and textile industries to a modern hub of medicine, research, and high-tech firms. Despite Durham’s rise over the past 10 years, non-Durhamites around the Research Triangle remember the Durham of old and are skeptical of it’s newfound success and reputation as a progressive yet gritty town.

The parallels between the rise of Durham’s revitalization and OpenStack’s popularity are uncanny. You still hear the following comments today:

“Why do you live in Durham, are you crazy?”
“How can you trust OpenStack community developers and run it in production?”

Enterprises continue to be skeptical of OpenStack’s production worthiness, but many companies are betting their businesses on this project. DreamHost, a Cumulus Linux customer, has been running a state-of-the-art OpenStack deployment for over two years. They automate their entire data center with Chef, leveraging Infrastructure as Code principles. Many others use standard DevOps Continue reading

Picking Up the Baton

I’m incredibly excited and honored to take on the role of CEO of Cumulus Networks. In many ways, I’ve trained for this role my whole life. I grew up in Silicon Valley. I have had a front row seat to the growth of the tech economy and been fortunate to watch many passionate leaders grow companies from simple concepts to multi-million dollar firms. I couldn’t be more committed than I am today to bringing a lifetime of experience and learning to bear in leading Cumulus Networks to its next phase.

First and foremost, thank you, JR, for entrusting me with this enormous responsibility. JR and Nolan have both invested their hearts, souls and many years of their lives in Cumulus Networks. They have hired incredible people, built great products, signed impactful partnerships and — in a brief few years — have already had a profound impact on this industry. They have fundamentally changed how networking products are bought, sold, developed and deployed, and in the process spawned a legion of imitators. I’m honored to be entrusted with the job of moving this organization forward. JR and I bring incredibly complementary skills to the table; he is a technical visionary and Continue reading

Finding Level

Nolan and I started Cumulus Networks with a specific vision: to help people build better, faster, easier networks.  To change the way that people think about building and deploying applications, regardless of scale. A lot of people have contributed into turning this vision into reality, and we’re excited by everything that we’ve achieved.

As we closed our series A, it was time to name a CEO, and we didn’t want to trust the company to a “professional CEO”. To that end, I took on the responsibility. In the early days I was able to stay involved with the technology and products; however, as the company has progressed, I’ve had less time to spend in the areas that motivated me to start the company.

Then along came Josh.  He participated in our extensive (some would say exhaustive) VP of Sales selection process and stood out.  His ability to grasp the business details as well as manage the team dynamics showed us that he has chops.  He joined us in June of 2015 and continued to impress.  He did his day job effectively by restructuring our sales team, refining the sales process, getting operations tight, and closing deals.  He also became a Continue reading

Fast & Furious with Mellanox

“Your data center is so cool!”  

That’s the common reaction to web-IT data center networks architected for data-intensive workloads and unprecedented agility. Clos designs optimized for east-west traffic flows and standard layer 3 protocols with ECMP have replaced brittle 3-tier Cisco-style layer 2 networks.  DevOps with a true Linux network operating system (AKA “NetDevOps”) enables secure, reliable configuration management and automated lifecycle management while converging on a common set of tools and processes across compute, storage, and networking. What’s really cool is the high speed interconnects in fat tree network designs.

10G and 40G leaf/spine networks were the standards last year. This year, 25G connected hosts and 100G spines are emerging as the desired interconnect, aligned with the latest Xeon-based servers with Broadwell cores. With the 25/50/100G inflection, many new merchant silicon entrants have hit the market, and Cisco announced proprietary ASICs for the Nexus 9200/9300 with availability at the end of the year. But, why wait?  25/50/100G Spectrum-based switches from Mellanox with Cumulus Linux are available now.    

The joint Mellanox Spectrum and Cumulus Linux solution enables an open platform, unlocked performance, and unleashed innovation. Cumulus Linux is based on an open framework enabling customers of all Continue reading

DevOps Tools for Modern Data Centers

Back in October, 2015, I spoke at All Things Open in Raleigh, North Carolina, an event focused on open technology and open source software. I was very excited by this event because many attendees work in or manage data centers, which means they are very familiar with Linux but have little experience with the networking stack. Cumulus Networks is the first major networking company to contribute a true Linux networking operating system for data center switches, which is highly disruptive to the industry and drives a lot of fun conversations with open-minded individuals.

The talk I did for All Things Open last October titled “Using DevOps Tools for Modern Data Centers” focuses on the new concept of NetDevOps or DevOps for Network devices. Since the network operating system is Cumulus Linux, why not use open source off-the-shelf automation tools that are already being leveraged in the data center to act as a controller.  These tools have an extremely large user base, are vendor neutral — that is, not proprietary — and can scale easily.

So what are the benefits of using open source tools? One of the most important benefits from a networking point of view is provisioning. Imagine you have 1000 Continue reading