For IT decision-makers, moving operations to the public cloud introduces the potential to rethink business processes from the ground up. For example, operating in the cloud means they can:
- Collaborate in ways not previously possible.
- Achieve productivity gains like never before.
- Transform the enormous amounts of data now available into artificial intelligence, machine learning, and other innovative applications considered science fiction just a few years ago.
But—and it’s a big but—these promises of the public cloud can come to pass only if the data is able to get where it needs to go, instantly, smoothly, and securely. And that means solving networking technology challenges that are specific to the public cloud environment.
At What Point Do Cloud Networking Technology Choices Matter?
As an organization makes initial forays into the public cloud, the choice of cloud networking technology doesn’t make much difference. Public cloud vendors’ native networking capabilities can comfortably handle relatively simple connectivity. The tipping point is when organizations expand their public cloud usage from a handful of virtual private clouds (VPCs) to hundreds of connections spanning multiple geographic regions and a range of public cloud providers.
Public cloud vendors’ built-in networking capabilities lack the range of features required to support increasing complexity.
At the same time, networking architectures with roots in the pre-cloud era—even virtualized versions—cannot deliver cloud-level security, performance, resilience, and other crucial capabilities. That’s because traditional networking architectures were built for data centers, which have static connectivity between physical resources, and not for cloud speed and flexibility. Data center networks must be provisioned to handle peak traffic volumes, even if those peaks occur in infrequent bursts. As a result, big networking pipes remain largely underutilized and organizations waste money supporting capacities that are rarely needed.
Additionally, traditional data center-based networking requires time, by highly trained and certified engineers, to establish or modify any connectivity. Cloud operations can’t deliver on cloud promises if an organization’s users need to wait weeks or months for networking
engineers to deploy a new application across a public cloud or hybrid cloud environment.
A New Paradigm for Cloud Networking
So, what should effective networking for the public cloud look like? Here are some guiding criteria to consider when choosing a modern cloud networking architecture:
Hierarchical. Cloud connectivity is not based on data movement between fixed, physical resources, so early cloud architectures tended to be flat rather than hierarchical. But hierarchical designs are crucial to enable things such as access control, reduced latency, and security for Internet of Things (IoT) and other distributed applications. Organizations operating key applications and resources in the public cloud must choose a networking architecture that is hierarchical, with tiers controlling core, distribution, and access functions.
Encrypted. In the public cloud, organizations no longer own—or can even locate—any of the physical hardware running the resources being connected. Ensuring the secure transmission of data, then, means depending on strong encryption and other security services that are built into the networking architecture itself.
Resilient. Because organizations don’t own the physical components of their cloud networks, their cloud networking technology must be resilient to both failures and constant infrastructure changes. In the old days, resiliency meant deploying another physical router as a failover. But who knows what physical equipment the public cloud vendor is using, or where it’s located? As with encryption, resiliency must be a fundamental property built into a cloud networking architecture.
Super fast. Mention “cloud speed,” and it’s understood to mean nearly no time between initiating and completing an action online. Traditional networking technologies are incapable of operating at cloud speed. Public cloud vendors’ native networking operates at cloud speed for connections among a handful of VPCs, but it can’t keep pace with more complex configurations. True cloud networking must be able to maintain super-fast cloud speed even as the complexity and number of connections continually increase and change.
Scalable. The public cloud represents tremendous compute power, which is expected to continue doubling every 18 months or so for the foreseeable future. Applications running in the cloud get smarter every year. Witness the progress with voice-enabled personal assistants such as Alexa and Google Home, or the rate of progress of technologies such as smartphones or IoT. Organizations need to select a cloud networking architecture that is able to scale in sync with the rapidly advancing cloud environment.
From Control to Fluidity
When organizations had physical datacenter-based networking equipment, everything was under their control. The price, however, was rigidity and inefficiency.
The public cloud offers extreme fluidity and flexibility, but the price is less control. While the top public cloud vendors—notably Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform—offer some networking capabilities, these cloud-native capabilities can always be stronger with more control enabled for the subscribing organization.
The answer lies in a new paradigm for cloud networking that leverages the public cloud vendors’ native networking features while adding the hierarchical design, encryption, resiliency, and raw speed needed to support a modern organization’s cloud operations—thereby forging a pathway to corporate success.
Dr. Sathaye is general partner of Cervin Ventures, and a board member of Aviatrix, a cloud networking company. Both organizations are based in Palo Alto, Calif.