Unicorns and the Cloud
Looking at SaaS companies that are considered Unicorns, I see similarities in that almost every entry is a cloud-native company. That is, they have deployed their applications and services within Cloud Service Providers (CSP) that include AWS, IBM Cloud, Google Cloud, and others. Separate from traditional enterprises transitioning to a cloud consumption model, this is another driver of strong growth, equal to 28% in 2015 and 69% for Amazon alone. And, even if a Unicorn outgrows cloud Platform-as-a-Service or Software-as-a-Service, they will still use the cloud for Infrastructure-as-a-Service rather than building out their own data centers (except in the rare instance of hyper-scale operators such as Google and Apple).
In the ‘legacy’ model, end-users traverse the Internet to applications hosted in enterprise data centers. An interim step includes the hybrid cloud and introduces WAN optimization and SD-WAN. But with the growth of the cloud, over-the-top media services, and e-commerce, this all changes. User traffic still travels from one ISP to another, but then jumps off into some CSP data center where the application and content reside. Given uncertainties in Internet connectivity, and the lack of any real control of what is termed the middle-mile, this isn’t the most efficient way to do business.
A Better Way
A more efficient and higher performance model shunts traffic onto the cloud as close to the end user as possible, leveraging CSPs for networking in the same way that we currently rely upon them for compute and storage. If the user’s application is hosted in the same country or region, great. There will still be some efficiency gains. However, the real benefits come into play if the content or application is in another region, and not necessarily hosted by the ‘onboard’ CSP. This content falls into three major buckets, spanning many verticals on the landscape: large or many files with personalized content, two-way real-time applications, and API-driven interactions.
Now focusing on Unicorns in more detail, a sampling of the verticals that can benefit from this new architecture more closely aligned to the cloud include:
· Content Collaboration and Management — Productivity gains via faster synchronization, better user experience drives service upsell, reduce number of CSP instances. For example, annual lost productivity costs in the US alone reach $7.5 billion.
· Marketing / Ecommerce — Productivity gains via faster synchronization and response, better user experience drives service upsell. This of course relates to on-line advertising as well, where longer load times leads to abandonment. In fact, at 4 seconds this approaches 25% and grows from there.
· Internet of Things — Reduce system response times for more accurate control including a feedback loop that supports cognitive computing, reduced times for large OTA updates.
· Online Media — Reduced latency and more predictable responses; reduce number of CSP instances. One useful metric is the Mean Opinion Score (MOS), where anything above 4 is considered good. In the absence of a more effective infrastructure, the score drops below this threshold in the absence of over-investment in resources.
· Games / Entertainment — deploy in new regions, better user experience
· Online Shopping — more effective ad performance, less abandonment
· Social Networking — better response for customized pages
With effective management tools, it is simpler to control performance end-to-end within CSPs than across the many ISPs, managing latency, throughput, and geography, while dynamically selecting different CSPs and even datacenters within a CSP based on real-time congestion or cost information. The ISP layer and its routing protocols is just not equipped to handle this dynamic, and in the same way that a Software Defined Networking approach is evolving for the data center and corporate WAN, the same ‘SD-Internet’ paradigm may now be applied across the globe.