The Birth Of The Distributed Cloud

There is a seismic shift that is about to reshape enterprise IT. Due to key trends like AI and IoT, applications and data as encapsulated in their supporting infrastructure are increasingly being spread across multiple datacenters – some on premises and some in multiple clouds – and edge sites.

This is a sharp demarcation from today’s IT environment, where organizations rely largely on traditional datacenters and individual clouds. According to Gartner, over half of enterprise-generated data will be produced and processed outside traditional datacenters or a single centralized cloud by 2022, compared to just 10 percent today. By 2025, they forecast that number will climb as high as 75 percent to 90 percent.

This evolution will give birth to two distinct and separate phenomena: The cloudifying of the edge and the rise of true multi-cloud. And these trends will eventually give rise to something even more radical: The distributed cloud.

Cloudifying The Edge

As noted above, some workloads are beginning to move to the edge rather than a traditional cloud. AI, ML, and IoT are currently the key technologies driving this change. These technologies increasingly leverage larger local datasets, which makes it necessary to process data right where it’s created rather than sending to the cloud and back.

Consider an AI/IoT use case like video surveillance that employs facial recognition. Use cases like this require low latency: the data needs to be analyzed locally so decisions can be made in real-time, as it is simply too slow to move this data to the cloud for processing – and because these data volumes are growing, it is also expensive.

People want to operate these edge deployments in a cloud-like fashion – they don’t want to manage them disparately across many different sites, like they have the past five to ten years when organizations were taking a siloed approach to managing tons of remote office sites. The coming effort to cloudify the edge means organizations will increasingly be able to deploy, support and connect a given app and its related infrastructure across various edge environments.

The cloudified edge will enable the following:

  • Fleet-wide management of distributed applications and data across heterogeneous infrastructure
  • Comprehensive and integrated compute, storage, networking and security for distributed edge locations
  • Secure, high-performance global connectivity across edge sites
  • Globally distributed control plane with Kubernetes APIs for application orchestration and multi-layer security for workloads and data

The movement to cloudify the edge is beginning, though it’s still in the early state. The big three cloud providers have all recently introduced new capabilities that bring a portion of their functionality to the edge in order to allow customers to begin managing edge sites in a more cloudlike fashion rather than as silos. This can be seen in Amazon Web Services Outposts, Google Cloud Anthos, and Microsoft Azure Arc. This is a good start, but these offerings only provide a limited set of functionality right now.

Multi-Cloud Over Multiple Clouds

It is important to note that this is a distinct and different trend from the move to cloudify the edge. But both feed into the evolution away from traditional datacenters and centralized clouds.

Many enterprises may claim they are multi-cloud today, but in reality they are just using multiple clouds individually and paying multiple cloud providers. That is akin to having two different server vendors or two different storage array vendors in an on-premises datacenter. These organizations typically only run each application in one single cloud provider (even if that application may be in multiple locations in that cloud). But true multi-cloud is a modular approach: for example, running component A of an app in Azure and component B in AWS. This true multi-cloud approach better embraces strengths of each cloud provider, allowing customers to leverage critical specialties of each cloud.

This strategy allows organizations to more easily support microservices and make those microservices more effective. Consider this analogy comparing an app to a family shopping for clothes. In a true multi-cloud approach, the mother can go to Saks to buy luxury clothing while the son can go to Patagonia to buy outdoor apparel, meeting differing needs. In a siloed approach similar to how most companies are using multiple clouds now, that family would simply be shopping at one single department store for all their needs.

Here is a real-world enterprise example. Imagine an app that has both a machine learning component and a Microsoft Office 365 database component. Google Cloud is strong for supporting ML, so that part of the app would go into GCP. Azure is naturally the fit to support the Office 365 database, so that component would go there.

True multi-cloud also yields better availability for each application (if one cloud goes down, you have the app in another) while also meeting certain compliance requirements (in case you need an application’s data to be located in a specific region). In a recent survey of CIOs, CTOs, enterprise architects and other IT execs, Propeller Insights found that 59 percent of respondents are putting the same apps in multiple clouds to maximize uptime and availability, while 54 percent do it for compliance reasons.

Looking Ahead: The Distributed Cloud

These early transitions to a cloudified edge and true multi-cloud are separate trends, but they are both part of a larger evolution toward the distributed cloud. The distributed cloud is an all new approach that will enable organizations to manage all disparate components – edge apps, apps stretched across multiple clouds, legacy datacenter apps, and the infrastructure that supports them all – as one logical cloud. This is the Holy Grail: being able to manage, operate and secure all these compute areas as a single, distributed cloud. That means deploying apps with a common set of policies and overarching visibility across all locations and heterogenous infrastructure.

The distributed cloud won’t happen overnight, and it really won’t start taking shape until 2022 and beyond. But these key trends illustrate that the first pieces are beginning to come together.

Ankur Singla is the founder and CEO of Volterra. Previously, he was the founder and CEO of Contrail Systems, which pioneered telco NFV and SDN technologies and was acquired by Juniper Networks in 2012. Prior to Contrail, Ankur was the CTO and VP Engineering at Aruba Networks, a global leader in wireless networking. Singla holds an MS in Electrical Engineering from Stanford University and a BS in Electrical Engineering from the University of Southern California.

Sign up to our Newsletter

Featuring highlights, analysis, and stories from the week directly from us to your inbox with nothing in between.
Subscribe now

2 Comments

  1. It seems that author missed to comment about Red Hat, which already supports seamless multicloud management. At the end of the day, you need a common interface to manage multiple services/instances/etc on multiple cloud or even on-premise / bare-metal. Dealing with all that is what Red Hat has been doing in recent years… check it out.

Leave a Reply

Your email address will not be published.


*


This site uses Akismet to reduce spam. Learn how your comment data is processed.