NOTE: This blog exists as part of a six part series.
The ongoing operational challenges associated with the multi cloud environment are well documented. As exemplified in a previous blog, the network and security design process is highly complex, involving multiple business partners, cloud service providers, colocation and data center space, and potentially traversing multiple bandwidth partner networks. Layered on top of that is the actual computing and application environment, which includes various user interfaces for each cloud provider to configure computing environments, different storage devices, networking services and security products. Finally, on top of it all are the business' overarching governance and policies associated with the system
The number of management systems and various panes of glass is simply overburdening existing networking, security, and IT resources. There are a few options at this point:
Clearly, the only solution that will address today's problems and pave the way for future digital transformation initiatives is the final solution listed above, deploying a new application-centric solution.
Clearly, the only solution that will address today's problems and pave the way for future digital transformation initiatives is the final solution listed above, deploying a new application-centric solution.
In the past few years, a few solutions have arisen that are worth looking into:
The traditional bandwidth carriers were slow to respond to the growing need for intercloud connectivity. Their existing products were too expensive or lacked the pricing flexibility (i.e., pricing elasticity) required for cost-effective cloud-to-cloud connectivity. Into the void rushed the current generation of middle-mile bandwidth brokers. These companies offer value in instantaneous layer two or layer three connectivity between cloud providers and colocation spaces, with pricing flexibility. These middle mile connectivity solutions don't solve all the multi cloud connectivity problems; however, they provide significant relief, as long as the business customer can meet these companies at one of their colocation POPs (as they do not extend connectivity over the last mile - thus the term “middle mile provider”). Today, these providers play an important role in “door to door” connectivity, establishing flexible bandwidth solutions between the cloud service provider networks.
Unfortunately, the upper layer issues of security, VPC integration, and application service level adjacencies still impede application level operational agility.
Next to market to resolve the multi cloud peering issue were first-generation Multi Cloud Networking (MCN) solutions. These providers leveraged the bandwidth solutions provided by middle mile providers, transit gateway capabilities, and open APIs offered by the major cloud providers to extend cloud-to-cloud network integration into each cloud provider's VPC. By integrating with cloud provider transit gateways, these MCN solutions could extend inter-cloud management “into” the cloud. Eventually, they released enhanced transit gateways of their own to offer additional flexibility. Transit gateways assisted with managing VPC integration within a region and across regions to other cloud providers by managing routing rulesets. The result was the ability to create networking connectivity between all cloud VPCs (not sites, but VPCs at each location) on a single pane of glass.
In the same way, first-generation solutions handle network connectivity; they also help simplify security by integrating with third-party providers or enabling their security software on the edge of their network to enable a ubiquitous security profile across multiple cloud environments.
Although first-generation multi cloud networking solutions took significant steps to help simplify multi cloud networking, there is still work to be done. As business customers are moving to the cloud, the private data center is not dead yet. To provide a full mesh capability between cloud providers and private data centers, those data centers must be full participants in the cloud connectivity fabric, not just stub networks. Additionally, carrier-class scalability is a must for carriers looking to deploy a solution to serve their customers.
Next-generation applications will be more intelligent, container-based, and further integrate advanced capabilities (artificial intelligence, process automation, microservice architectures, and quantum computing) into their operations. Businesses continue to look for improvements in application performance. If the network is to support these capabilities, any multi cloud solution needs to become application aware now, not later. Unfortunately, “Application-Aware” means different things to different people, and the term has become a bit “cloudy.” In this case, it means instantaneously being aware of explicit application workload needs at the service level while ensuring service levels are met in delivering that connectivity.
Edge Cloud providers are expanding, and applications will be further distributed in the coming years. The cloud edge ecosystem will be even more distributed that the existing cloud model, with many providers serving many differing requirements. It will be imperative to integrate these providers into any solution, and virtually erase the boundaries across edge cloud providers and between the hyperscalers and the edge cloud providers. Edge Cloud providers are expanding, and applications will be further distributed in the coming years. The cloud edge ecosystem will be even more distributed that the existing cloud model, with many providers serving many differing requirements. It will be imperative to integrate these providers into any solution, and virtually erase the boundaries across edge cloud providers and between the hyperscalers and the edge cloud providers.
It is essential to recognize the advances made in this networking space over the past few years, while simultaneously recognizing that work remains to bring the application layer into total synchronization with the network layer that supports it. The complete integration of application workloads across a diverse underlying network infrastructure is a must to meet future goals, and the planning, testing, and process understanding needs to start immediately, or the prospect of falling behind with the next wave of cloud computing becomes certain.
