Cumulus® Linux®

Unleashing the Power of Open Networking

Choose the hardware, software, and management and automation tools that best fit your application needs. Cumulus Linux is supported on an ever-increasing number of industry-standard platforms.

Cumulus Architecture

  • Automation Open or Close

    Cumulus Linux is the first full-featured Linux Networking OS for modern data centers. A key ingredient to a modern data center is automation. Automation is easy to set up and implement with Cumulus Linux, allowing you to take advantage of the latest Linux applications and automation tools that have been serving the compute world for years. By extending a standardized automation toolset from your compute environment into the networking world, you can automate the complete operational life cycle of network devices from configuration and provisioning, to policy-based change management.

    One of the key benefits of Cumulus Linux is the increase in network stability and uptime. This is done by preventing inconsistencies and misconfigurations driven by inflexible vendor locked-in systems. Manual approaches to working with complex protocols are no longer needed. Now you can choose from a number of native Linux applications, in addition to third-party open source and commercial applications, which can be integrated as add-on Linux packages to optimize your business. Cumulus Linux enables a consistent experience between the network and compute, driving the next wave of scale, collaboration and innovation in networking. Try any of the automation toolsets to reduce costs by automating your network and delivering applications with higher SLAs and vastly reduced deployment times.

  • OpenStack Open or Close

    OpenStack is all about making the data center easy to deploy, configure, and scale, while keeping it economical. To build OpenStack clusters, customers continue to take advantage of the open compute and storage standards but until recently networking was the bottleneck — it was isolated, and hard for linux system administrators to learn. Due to traditional networking vendors using proprietary vendor-specific software, OpenStack clusters were only as flexible as the Top of Rack (ToR) switch.

    Cumulus Networks native Linux Network OS (NOS) makes it easy and economical to deploy, configure and manage because the entire stack now runs Linux in the entire rack. The data plane and control plane are now “speaking the same language” without complex compute APIs or networking CLIs.

    With common Linux development and DevOps tools, it's simple to collaborate across technology silos with wide Linux expertise and build pre-packaged automation templates/workflows for implementing best practices and minimizing transition time.

  • Converged Infrastructure Open or Close

    Enterprise IT infrastructure is evolving from a legacy, vendor-driven, three-tiered approach to a Web-scale IT model that relies on software. Storage-based software solutions manage flexible x86-based scale-out hardware resources that bundle compute, memory, and storage into each node.

    Hyper-converged Infrastructure is a transformational approach that combines simplicity and scale-out by eliminating dedicated storage nodes, yet provides data resiliency via distributed storage software. By coupling Cumulus Linux — the network OS for the open networking movement — with hyper-converged systems, enterprises can break free from the traditional siloed approach and can simplify and aggregate compute and storage resources.

    Storage is a critical component of the converted data center architecture as applications are designed to scale. One can deliver server and storage resources that are preconfigured to work together, thus minimizing the risks, creating a predictable cost model. With the evolution of intelligence in the software-defined data center and leveraging hardware off the shelf, Cumulus Networks can align with storage partners that have a shared vision of presenting turnkey solutions to customers.

    The benefits of using Hyper-converged Infrastructure are very complementary to the benefits of using Cumulus Linux for networking. Combined with Cumulus Linux-based scale-out, the entire networking stack follows the same design principles as compute to focus on scale-out, simplicity, and automation readiness. The consumption model of an integrated stack through “Data Center in a Rack” can be easily driven through the hyper-converged solutions with open networking to help enterprise customers deploy Web-scale architectures at ease.

    Converged infrastructure can serve as an enabling platform for private and public cloud computing services, including Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and Software as a Service (SaaS) offerings.

  • Monitoring & Analytics Open or Close

    Modern Data centers are are a Grand Central of information. Network Architects designing Data Centers on web-IT methodology would prefer to have the ability to monitor everything and have a self -healing proactive network, resulting in a simple, converged and controlled environment. Cumulus Linux's open framework creates a very transparent and easy platform for partners and customers to leverage third party agents, build extensions, make enhancements and contribute them back to the community.

    Cumulus Linux allows Network Operators to always have context (who, where, when) for anything new that arises, and understands the types information to gather and analyze for better design. Convergence through use of existing tools extending with Cumulus Linux provides the standardization across the compute and network teams for operational efficiency. Finally, having 100% visibility is a fundamental prerequisite for closed-loop control execution in any infrastructure operations center.

  • VMware Open or Close

    In the last decade, VMware transformed the compute environment with virtualization, which later extended beyond servers to include storage, application and now network virtualization. VMware allowed disaggregation of server hardware and software that radically improved provisioning speed and economics.

    With Cumulus Linux and disaggregation of software from the network hardware (a.k.a Open Networking), speed and economy of the data center network and applications running on them have dramatically improved.

    At the core of this new approach is the Cumulus Linux network operating system, offering the industry’s first true native Linux experience for networking. Cumulus Linux combines modern day software and OS architectures, an open framework for development with Linux and a real-time programmatic, seamless and automated model for cloud networking.

    Take a look at the joint solutions between VMware and Cumulus Networks and how it can help you through the transition from legacy networking.

  • Network Virtualization Open or Close

    Cumulus Linux drives the bare metal networking paradigm by leveraging industry standard hardware and the advancements in merchant silicon. With the interest in overlays and customers demanding workload mobility, a centralized pane of glass for managing virtualized and non-virtualized workloads has become essential. Software overlays have been unable to address all workloads and the limited visibility from overlays into physical networks make it hard to provision, automate and troubleshoot networks.

    Cumulus Linux can act as the hardware L2 gateway for enabling the high performance desired for workloads while leveraging overlays, enforcing policies as desired and scale with the benefits of VXLAN technology. The un-tethering of the switch hardware from the operating system software provides the foundation for a software-defined data center. By having the intelligence abstracted in the software and all the change management occurring in the logical layer, Cumulus Linux is able to drive tremendous data center economics with end-to-end operational efficiencies.

  • Big Data Open or Close

    Hadoop is one of the most disruptive innovations in enterprise IT, and its use is extending from high-performance computing and research organizations to enterprises setting up cloud environments. Hadoop is an open source project that offers a powerful distributed platform to store and manage big data. Hadoop Distributed File System (HDFS) is designed to be a scalable, fault-tolerant, distributed storage system that works closely with MapReduce on commodity servers. Cumulus Networks shares this vision of having network infrastructure run on standard open networking hardware, and supports the Hadoop environment with Cumulus Linux, the networking operating system for open networking.

    With multi-tier scale-out IP Clos fabric designs supported through Cumulus Linux, the network infrastructure enables large data sets to be transferred and processed rapidly, thus making it ideal for clustering applications such as Hadoop.

    Specific features like rack awareness, minimal data motion, utilities, rollback and Standby NameNode ensure that the Hadoop clusters are highly functional and highly available; in addition, they are enhanced by several Cumulus Linux features.

    Manageability has become a key expectation from Hadoop users. Hadoop has made great advances with Ambari, which enables the automation of initial installation, rolling upgrades without service disruption, high availability and disaster recovery — all of which are critical to efficient IT operations.

    With Cumulus Linux, there are a variety of toolsets you can leverage for automating the complete infrastructure stack through zero touch provisioning, thus integrating easily into the user models of typical Hadoop deployments.

Cumulus® Linux® Architecture

The modern data center is the supercomputer reinvented. Some call it warehouse computing for the sheer scale of the systems.

The workhorse for host networking in modern workloads, Linux continuously evolves to address bare metal, virtual machine, and container deployments with rich functionality. If you think you’re not using Linux networking already, think again. Cumulus® Linux® preserves that goodness by leveraging the latest in networking silicon to hardware accelerate the native Linux data path for switching use-cases.

This implementation presents the high speed switching ports just like a NIC, preserving the Linux abstractions for user space applications and networking constructs such as interfaces, bonding, bridging, routing, tunneling, mpls, and vrf. A device driver, aptly named “switchd”, acts as a conduit between the kernel and the underlying networking silicon. It’s that simple.

No network switch is complete without the set of protocols that allow it to play nicely with its peers. The pure Linux API affords choice amongst a broad range of protocol offerings; we started with a set that provided a good mix of function and maturity and enhance them with improved performance and relevant features. You can use the world class protocols that come with Cumulus Linux or pick your own journey without fear. As an example, we’ve spent a ton of development effort hot-rodding the Quagga routing suite, yet we have a few customers that have written special purpose routing protocols.

The Linux operations eco-system has a rich and ever-evolving set of tools, de facto data exchange formats, and platforms for deployment, configuration, monitoring, troubleshooting, and orchestration. Again, the pure-play Linux system allows these mechanisms to be easily leveraged in Cumulus Linux; native Nagios/Sensu checkers Puppet, Chef, Ansible, and Salt configuration modules; and log management frameworks can all be used out of the box. As always, standard usage of the Linux filesystems like /etc, /sys, and /proc means that any extensions are a script away.

This purity-of-approach to Linux networking enables Linux system tools to exist unmodified in Cumulus Linux; which is not true of most of our industry peers. This means that enhancements, bug fixes, and most importantly security updates are available to you as soon as they are published.

In today’s webscale-IT world, you have to ask yourself “why NOT Linux”; we did that and Cumulus Linux is our answer.