Abstracting Your Way to a Multi-Cloud Network
The enterprise is on a course to a multi-cloud architecture, which may come as a surprise to some top-level executives who still aren’t fully aware of what their underlings are up to. While this will ultimately make the data environment more resilient and deliver better service to users, it throws a wrench into plans for streamlined network infrastructure.
Fortunately, there are few connectivity problems that cannot be solved through abstraction. But in order to achieve even a rudimentary multi-cloud architecture, the enterprise will need to figure out how to integrate virtual networks into the LAN, the wide area and carrier levels.
According to a survey reported by Business Cloud News, less than half of organizations have a multi-cloud strategy, while upwards of a third do not even have plans in place for single public or private clouds. This is dangerous because as big data and the Internet of Things (IoT) emerge as core business assets, a disorganized cloud will become a key limiting factor to the delivery of digital services. Whether the task at hand is data migration, compliance, backup and recovery or performance optimization, the common denominator is the network’s ability to share and exchange data.
Only recently, however, have the top networking vendors turned their attention to optimizing their virtual architectures for multi-cloud environments. Cisco recently announced plans for field trials of its Application Centric Infrastructure (ACI) Virtual Edge component, through which it intends to integrate connectivity among the three major clouds: Amazon Web Services, Google Cloud Platform and Microsoft Azure. If successful, the platform will allow organizations to run applications on their private and public deployments under a unified set of network policies, including the all-important task of load-balancing across distributed, disparate infrastructure.
Meanwhile, VMware is turning its attention to virtual networking in multi-cloud environments through updates to its VMware Cloud and NSX platforms. Recent additions include discovery and cost management tools that provide visibility into hybrid clouds in support of resource utilization and aggregation, along with workload distribution and the ability to pool resources across independent clouds. As well, a new network insight module provides traffic flow analysis and troubleshooting, while the NSX Cloud service offers microsegmentation of security and other functions so they can follow applications wherever they are deployed.
Of course, coordinated networking on the cloud layer is only as useful as the underlying connectivity between clouds, which is why many organizations are looking for ways to virtualize their wide area infrastructure as well. Under a coordinated SD-WAN approach, the enterprise can establish a single management pane to view and program their network, says Riverbed CTO Hansang Bae in an interview with Silicon Angle. In this way, all data points become equal regardless of where they are hosted, and they can still be grouped according to their function, say, as an IoT point or a backend services point.
While a multi-cloud future seems inevitable for most organizations, it will still be a far cry from the self-managing, dynamically optimized environment that made its way into the hype cycle of the early cloud era. Abstract networking across public, private and hybrid infrastructure removes many of the roadblocks that would otherwise hamper data performance, but the enterprise will still have to deal with constant issues surrounding governance, security, connectivity and the like.
At the moment, the foundational elements of the hybrid cloud are in place, but the higher level details will only be solved through extensive trial and error.
Arthur Cole is a freelance journalist with more than 25 years’ experience covering enterprise IT, telecommunications and other high-tech industries.