Networks just keep growing, don’t they? They’ve evolved from a few machines on a LAN to the introduction of Wi-Fi—and with the Internet of Things (IoT), we’ve now got a whole new class of devices. Throw in the rise of smartphones and tablets, cloud and edge computing, and network management starts to get a little unwieldy. Managing a network with 300 devices manually might be possible—300,000 devices, not so much.
What is web-scale automation?
Network automation has been around awhile now, in various names from various vendors, using a number of proprietary protocols. The key word being “proprietary.” Many traditional network vendors design a well-functioning network automation system, but participate in vendor lock-in by ensuring that the associated automation stack, and its requisite protocols, only run on their hardware.
Web-scale automation is different. It relies on open, extendable standards like HTTPS, JSON, and netconf, among an ever-increasing number of systems and solutions. With web-scale automation in your organization, network management can over time become a background function; something that only notifies you in exceptional circumstances.
This does not, in any way, reduce the need for those who know networks to be employed at your organization—it simply reduces the amount of work needed to maintain and manage ever-expanding network architectures.
How can web-scale automation help my organization directly?
In the past, a switch was a switch, a router was a router and a server was a server. And for a while, that worked. Different people—different teams—were set up to manage these things, and all worked well. However, that approach no longer works, because virtualization and containerization increase workload density. Added to that burden is edge computing, which increases the complexity of networking hardware and software, and you can see how the problem becomes untenable.
Web-scale automation allows you to manage switches, routers and network functions virtualization (NFV) services in the same way—and with the same DevOps tools—you would use to manage a server. This gives rise to the term “NetDevOps,” a combination of previously separate networking, development and operations.
With unified, single-style web-scale automation, your router, NFV services, switches and servers all talk the same way, using the same protocols, all the time. From initial boot up and configuration, to hands-off, instant and wide-ranging policy changes, web-scale automation is the only viable way to implement management decisions at scale.
That brings us to Cumulus Linux. Cumulus Linux is built on a standardized, Debian-based core. Debian and Ubuntu are popular server operating systems, and much of the management that works on a server running either of these operating systems will also work on a Cumulus switch. Standardized technologies bring massive efficiency gains to any datacenter, keeping your service-level agreements (SLAs) high, your deployment time low, and your errors and downtime minimal.
There are many networking technologies in play today, from many vendors. They all claim, especially now, to be automated. To be hands-off. But are they really?
In some cases, true network orchestration—the automation of the numerous different individual automations necessary to manage at scale—is only possible if an organization buys all their networking (both physical and virtual) from the same vendor. Unfortunately, even when organizations do commit to single-sourcing their networking, the network orchestration available leaves much to be desired.
The industry-leading vendor may have a wide selection of products available. But that catalog was built through acquisition, and the result is a single vendor shipping a double-digit number of operating systems, with features, capabilities, and even licensing schemes that vary wildly between offerings. Viewed through this lens it’s easier to understand why vendors with such broad offering diversity (and market dominance) have failed to provide compelling software-defined networking (SDN), and still struggle with network orchestration and management.
Cumulus Linux, with its standardized, datacenter-style automation and provisioning models, provides huge gains in productivity, while greatly decreasing downtime, which is necessary to function in tomorrow’s world. As artificial intelligence, IoT, 5G and SD-WAN technologies explode across all industries, larger vendors need to keep up, or they’ll rapidly become irrelevant.
In a world dominated by the relentless rise of always-on, high-speed network connectivity, automation, orchestration and management are essential—from the datacenter to a remote sheep farm, and everywhere in between.