Predictions for Cloud Computing - 2017

Now that 2016 is almost over, I wanted to sit down and transcribe the trends I'm seeing in the field including the strategies undertaken by various companies.

Enterprise Cloud adoption is set to accelerate - perhaps a tame one to start with, but until 2016, most large enterprises had either experimented with private, on-premises cloud or private off-premises cloud(s). In both cases, the only workloads targeted were dev/test. There were some who reclassified their VMware farm as 'private cloud'. Outsourcing providers who offered IT services and performed tasks such as capacity planning and on-demand provisioning relabeled themselves 'cloud like'. Discounting those, it was only a few outliers or early movers in each industry which were the ones with sizable investment (capital and workload) in a cloud environment that wasn't VMware based. The inhibitors to adoption of cloud computing are (in the order of severity) governance (or the desire to control workloads, also known as the desire for the traditional IT departments to be brokers of IT even though the provisioned infrastructure may not be within the four walls of the corporate data centre), risk mitigation (not to be confused with security, though it is certainly a part of it) strategies to provide a layer of safety (typically, this is performed by establishing an Enterprise Architecture or Cloud Architecture board which decides where a given workload should go - be it on-premises, cloud provider A or cloud provider B, etc), security, networking, and ability to dictate requirements. It is the last item which has been the most difficult to grasp for enterprises since the megacloud providers are generally not swayed by a single large client's request to add new features and the sense of losing control has dissuaded the executives in moving quickly with cloud adoption.

SDN and SD-WAN - The road to SDN started over 20 years ago when Java was released, with AT&T GeoPlex being one of the world's first SDN projects. Beyond an 'active network' or a 'network API', SDN advanced rapidly with advances in OpenFlow and Network OS (circa 2008/2009 with OpenFlow v1.1 release in 2011), and availability of vSwitch about the same time. Lack of SDN capabilities at the enterprise and therefore the ability to carry over existing network segments has been a big inhibitor. Most cloud providers now have SDN at the data center (though in various stages of maturity) - AWS allows this with VPC subnets, Azure with VNets and IBM with the ADN feature (in beta). Cisco's ACI and VMware's NSX are the predominant tools for SDN in the data center, and one of the easiest plays for enterprise cloud adoption is the use of these SDN technologies to extend the VMware farm into a cloud provider. This is exactly the reason why VMware has formed partnerships with IBM and Amazon. Software-defined wide area network (SD-WAN) is a specific application of software-defined networking (SDN) technology applied to WAN connections, which are used to connect enterprise networks – including branch offices and data centers – over large geographic distances. SD-WAN can be used to deliver enterprise services such as virtual private networks (VPNs), WAN acceleration, and bandwidth optimization. By using commodity off-the-shelf (COTS) in branch offices, SD-WAN as a service using orchestration can be delivered with the aid of virtual customer premises (vCPE) or lightweight CPE equipment. Cisco's iWAN and Citrix NetScaler are two key offerings in this space. NetScaler partners with Amazon while IBM has their own MultiNetwork Services.

The rise of Containers - The internet is full of claims of Container-as-a-Service (CaaS) by Docker, The New Stack, and plenty others. The rise of Docker, Open Container Initiative, the open source cluster manager Mesosphere (which has evolved to organize the machines in your datacenter or cloud as if they were one big computer), support for Docker by all megacloud providers, and convergence of container runtime between Docker and CloudFoundry in CloudFoundry Diego have all resulted in faster adoption of Containers. Add to that the rise of container orchestration tools and this is what has further accelerated adoption - Kubernetes for automating deployment, scaling, and management of containerized applications, Mesosphere's Marathon as an orchestration framework for management of long-running applications, Docker Datacenter which delivers an integrated platform for developers and IT operations to collaborate in the enterprise software supply chain, etc. In 2016, I met the executives from at least a dozen enterprises who were in various stages of experimentation with containers but only a couple of applications in production. I believe that will change drastically in 2017.

Partnerships - If you have following so far, this has been covered already in a way. Historically, Amazon, Google and Microsoft went after scale based on here, here and here; a good article on Microsoft's DC architecture is here), while IBM chose to go global with over 40 locations worldwide to provide data sovereignty in more geographies than others. That said, Moore's law's has been ruthlessly impacting cloud computing costs (some say cloud computing is even beating Moore's law, such as The Economist, and Nimbix, but it hasn't been around long enough in my opinion to provide a reasonable justification), the utility computing price the megacloud providers are all struggling for differentiation. Any new feature introduced by any one of the cloud providers will be quickly copied by the rest very quickly (AWS object storage introduced by Google, IBM and MS; IBM private / dedicated servers introduced by AWS, etc). Even if you one provider is furthest ahead in the field, the differentiating features can be copied quickly, so the next best option is to forget partnerships. In 2016, we saw IBM partner with VMware, VMware with AWS, Microsoft with Adobe and GE, among others. I foresee the partnerships accelerating especially in the application and DevOps space.

Consolidation - Getting a cloud strategy right for any IT vendor is challenging, especially since all established giants run headfirst into innovator's dilemma and perhaps more painfully, into Conway's law in terms of their offerings and user experience. HP shut down its Helion public cloud in Jan 2016, Cisco announced intentions to shut down it's Intercloud, while VMware its vCloud air strategy to partnerships (as covered above). I expect the trend to continue despite the likes of SAP and Oracle venturing out build their own clouds. shift its vCloud air strategy to partnerships (as covered above). I expect the trend to continue despite the likes of SAP and Oracle venturing out build their own clouds.

The focus on PaaS and Apps - In the first decade of cloud, the focus was heavily on infrastructure - obtaining parity with all the infrastructure elements required to host applications in someone else's data center. Arguably the features / offerings can be optimized, especially in networking, but the bulk of the focus will be on apps. I talked about containers already, and with CloudFoundry, apps are ever more independent of the underlying infrastructure. With Diego that independence continues and CloudFoundry gains parity with the Docker runtime. AWS introduced serverless computing with Lambda, and IBM followed up with OpenWhisk. The next battle will be for differentiated application services that provide 'stickiness' - reasons to use and stay with a given cloud. This is where IBM's position on cognitive computing may be the reason developers flock to Bluemix, or to Google for Cloud Machine Learning. Microsoft is the only vendor that has a gravity field around it due to MS apps and the work done on Azure to provide native support for MS libraries. Though there are some chinks in their armor due to peer pressure and with support on .NET 5.0 in CloudFoundry and the support of SQL server on Linux. I expect more features for application support to be introduced in 2017.