Cloud-native networking is all about providing access to the never-ending pools of resources that now reside in the cloud. While we hear the term cloud computing pretty often and know what it means, cloud-native networking sometimes sounds unfamiliar at best. This is because while most of us are used to the network appearing out of thin air to provide connectivity, what we don’t see are the wheels turning behind the curtains that make this possible. Signing up to use cloud resources without properly planning for the required network demands is a lot like building a city without roads. Several organizations have gone down this road already and figured out only too late that neglecting networking resources is a critical mistake. Cloud-native networking provides users with the proper network capacity to fully make use of the clouds features and benefits, without which, migrating to the cloud is pretty pointless.
According to IDC, 90 percent of enterprises will use multiple cloud services and platforms by 2021. The sheer network infrastructure that’s going to be needed to support communications at that scale is mind-numbing just to think about. While public cloud providers do provide native tools for hybrid cloud setups that are pretty straightforward to use, these solutions tend to get harder to scale up as organizations grow. AWS, for example, offers a dedicated connectivity solution called AWS Direct Connect that allows customers to establish a dedicated network connection from their on-premises location to the cloud without using the public internet. In addition to quite often reducing costs, this also increases performance, security, and bandwidth! Equinix, one of their program partners, just announced participation in the new AWS Direct Connect Service Delivery Program that enables their customers to use hosted connections with 1G, 2G, 5G, and 10G capabilities on ECX Fabric.
Similarly, Google has Interconnect and Azure has an express route, in addition to the newly announced Microsoft Azure Networking MSP Program, a push to improve connectivity into their environments. According to Yousef Khalidi, corporate vice president of Azure Networking for Microsoft, the new program aims at enabling other partners and network carriers to leverage their own networking capabilities with Azure’s own expanding range of networking tools. There are also several independent network-as-a service providers like Datapath.io, Aviatrix. Masergy and Aryaka, and these actually tend to be easier to scale and run across clouds since they’re not tied to any particular public cloud. NS1, the leader in next-generation DNS and traffic management solutions, just announced the acquisition of network performance optimization technologies from Datapath.io. NS1 customers will be able to take advantage of the “always-on” optimization as well as the dedicated “full-mesh” connectivity across cloud regions and providers immediately, with no additional setup or configuration.
While WAN is a popular way to access resources in the cloud, the migration to cloud-based applications creates an ever-increasing demand on bandwidth. A demand that traditional enterprise WAN solutions are finding it very hard to meet, competitively. Now, when cloud-native networking solutions are used to manage and deploy functions across the WAN, it’s often described as software-defined WAN (SD-WAN) or as some call it, cloud WAN. According to the Frost & Sullivan global enterprise SD-WAN survey of 2018, 94 percent of enterprises are either considering or have already deployed SD-WAN solutions to address evolving WAN requirements. Those are pretty big numbers and enough evidence that traditional WAN just doesn’t cut it at his scale. When it comes to networking, not unlike everything else that’s software-defined, the ultimate goal is to make it so that services are agnostic to infrastructure, only then can communications happen swiftly across networks and providers.
At the moment, the SD-WAN infrastructure market is pretty much dominated by Cisco and VMware, with the former taking an early lead with its acquisition of startup rival Viptela two years ago. This was when Cisco already had its own IWAN and Meraki SD-WAN solutions. VMware SD-WAN is made up of a distributed network of VMware SD-WAN Gateways, a cloud-based VMware SD-WAN Orchestrator, and a branch platform called VMware SD-WAN Edge. Other popular providers include Juniper networks, Citrix, and Silver Peak. Microsoft is on the cloud networking hunt with the aforementioned MSP program and has selected SD-WAN provider Aryaka to be its first partner. The new offering uses a combination of SD-WAN technology and a global managed service to help speed up and simplify SD-WAN connectivity to Azure. Additionally, while Aryaka is also providing Azure with WAN monitoring through the MyAryaka portal, it’s also partnered with rival cloud AWS with Aryaka SmartConnect SD-WAN for AWS.
With cloud-native networking, in particular, you will find the word “fabric” used quite often as a metaphor to describe the complexity of network topology. While the term has no doubt been derived from “switching fabric” that refers to the switching units and the integrated circuits contained in a node, the term “cloud fabric” is used in collective reference to every piece of equipment holding the network together. Incidentally, Facebook has named its data center plan “data center fabric” and has even built a fabric aggregator to help the social media giant keep up with traffic. With regards to switches, in particular, the most common problems occur on a physical level where cabling is virtually impossible considering the total number of connections and intense traffic flow. This is why almost all modern centers (including Facebook) now use fixed-port switches in leaf/spine topologies.
All this fabric flying around may seem confusing at first, the basic point, however, is that it’s more than the human mind can wrap itself around. Have you ever tried counting the threads in a piece of cloth? Pluribus Networks, a leader in next-generation software-defined networking, announced earlier this year that they were taking two of their products, Netvisor ONE OS and Adaptive Cloud Fabric, and extending them to support emerging edge computing and distributed cloud architectures. While this may seem like an insignificant piece of news to some, it’s quite indicative of a growing trend where edge finally meets cloud. As the IoT continues to explode and new technologies like 5G emerge, resources are going to have to exist a lot closer to the edge to enable the levels of service people expect. Big Switch Networks is another noteworthy mention in this field with Big Cloud Fabric that features SDN controllers coupled to “white-box” hardware switches.
Cloud-native networking: Assortment of parts and platforms
Modern cloud-native networking is all about integrations between an assortment of parts and platforms, across a variety of clouds and physical locations. This means in addition to being quick, reliable and everything else we expect as demanding customers, modern cloud-native networks need to be exceptionally elastic. They also need to be future proof and built to be compatible with technologies that haven’t arrived yet. While everyone and their grandmothers are banking big on multi-cloud networking, SDN, SD-WAN, and the IoT, it’s important to keep track of the hardware aspect of all this “virtual” networking. Physical connections don’t magically disappear and the latest SDN technologies often require the latest network fabric hardware to make it all work. Organizations migrating to the cloud usually neglect one aspect or the other.
Featured image: Flickr / Marc Veraart