MWC 2017: IoT, edge and cloud - friends or foes?
We have seen all kinds of 'things' – from fitness trackers to surveillance drones and smart light poles – being showcased at Mobile World Congress 2017. These 'things' are meant to collect all sorts of data. But where does all this data get processed – at the edge or in the cloud? Respondents to 451 Research's Voice of the Enterprise IoT Workloads and Key Projects 2016 study indicated that current applications are being analyzed on IT infrastructure located where the data is generated (44%) or on company owned/leased datacenter facilities local to the edge (55%) far more frequently than they are sent off-site to IaaS/PaaS or SaaS. However, over time, we expect more of this analysis to shift to the cloud.
The 451 Take
Ultimately, intelligent edge computing and gateways will not replace cloud computing (despite the hyperbole from blockchain enthusiasts), nor will the cloud somehow supplant the demands of edge computation. The IoT market decades from today will be a multipolar one, with the continuum of compute, analysis and storage envisioned by the OpenFog Consortium with some applications processed on or near the originating source; some on-premises due to technical, security or regulatory constraints; some nearby in operator networks (mobile edge computing); at interconnect sites by the Equinixes and EdgeConnexes of the world; and in public and private clouds. What was evident from MWC 2017 was that the IT industry is in a period of discovery about the diversity of IoT/OT use cases and applications. Some of these 'newly discovered' legacy use cases will require retooling of products, services and business models, the latter being the most challenging for large IT technology vendors, as well as the buildout of partner ecosystems to address the multitude of market requirements that cannot be met with 'off the shelf' technology. If there was one underlying IoT theme at MWC 2017, it was the widespread partnerships between technology titans and startups alike.
Edge computing encompasses data analyzed on the originating device or near-edge on a computing device or gateway. Processing data at the edge instead of the cloud or a data warehouse makes sense, for instance, in applications such as precision manufacturing and transportation where decisions need to be made quickly and locally, or when bandwidth availability or cost is an issue. IoT applications with stringent latency demands put more emphasis on local computation. Edge computing has the ability to reduce latency (no need to send the data to the cloud and wait for a response) and address potential network bottlenecks. However, the computing power and storage capacity of many devices are limited, and not all devices that need to work together are smart enough.
Edge computing was pervasive at MWC: robust compute and sensor capability from Intel, ARM Holdings, Samsung and HARMAN, and operating systems to enable these edge capabilities from companies including Ubuntu (Canonical) and Red Hat. There was a lot of IoT M&A activity – 44 deals – surrounding edge computing in 2015 and 2016 (excluding large semiconductor deals such as Qualcomm's acquisition of NXP Semiconductors and SoftBank's acquisition of ARM, taking it private).
Gateways sit next to endpoints, both IP-native devices and legacy endpoints, and perform the physical and protocol translations required to extract data for analysis by upstream applications, on the gateway or in the cloud. The gateway market has grown rapidly, as demonstrated by the sheer quantity of IoT gateway platforms on display at MWC. Dell Technologies leveraged the show to launch a new gateway to round out the lower end of its gateway portfolio, while HPE was demonstrating its EdgeLine family of gateways that also includes larger models with integrated compute and analog I/O capability for more demanding local analysis. In June 2016, Dell sibling VMware launched Liota (Little IoT Agent), an open source software development kit for building secure IoT gateway data and control orchestration applications.
For all the 'intelligent edge' messaging at the show, the importance of the cloud was not overlooked. The breadth of the IoT market, which is an abstraction of hundreds of distinctly separate markets in itself, means that not all data generated will have the stringent latency demands of an electrical substation or semi-truck brake pedal. Outside of these 'control loop' applications, the majority of value from IoT will be derived from the aggregation and integration of operational sensor data with business systems, which will happen in the cloud.
Cloud platforms can handle a high number of connections and data points and run complex jobs simultaneously, such as big-data analytics. The cloud acts as the connection, storage and analytics hub, and in many cases, the connected devices don't need to be smart because the cloud provides all the required intelligence.
Market leaders Microsoft (Azure IoT) and Amazon (AWS IoT) were balancing their cloud messages with nods to the intelligent edge at the show, as were IBM, Oracle and Cloudera. Communications PaaS provider Twilio has partnered with T-Mobile and launched 'Programmable Wireless,' a SIM card that provides data, SMS and voice communications to connected devices. Powered by its cloud, Twilio will deliver capabilities (including a new device API) that support developers in building cellular-connected offerings.
What are cloud providers doing?
IoT presents cloud providers with an opportunity they can't ignore: billions of connected devices exchanging huge amounts of data that needs to be stored, processed and analyzed. The leaders in the IoT cloud platform provider segment, according to customer responses in the most recent 451 Research Voice of the Enterprise IoT quantitative study, are Microsoft, AWS and Google. As stated above, providers are walking a fine line between legacy IT 'cloud first' strategies and enabling the intelligent edge use cases in markets such as industrial IoT and transportation that constitute the majority of IoT project spending today.
Verizon has bet on a combination of M&A, organic development in IoT platforms and people, ecosystem building, and networks. The company has been looking for new growth areas around connected smart devices, and it recently completed its acquisition of Sensity. The company's IoT development platform, ThingSpace, provides a data platform of record that can unify deployment silos within cities and also within Verizon. This is part of Verizon's smart cities initiative, which is in trial mode in Boston.
VMware's route to IoT is via strategic alliances, where it expects to leverage its strengths in device management, operational analytics and security to ultimately help partners and customers extend their reach from the datacenter and cloud to the edge. At MWC 2017, VMware and HARMAN announced a partnership to deliver IoT offerings to automotive, retail, industrial, building management and energy-efficiency industries. Other VMware partners include Bayshore Networks, Dell, Intwine Connect, Deloitte Digital, PTC and V5 Systems.
The OpenStack Foundation is running user surveys on IoT to get a better understanding of potential requirements, but we haven't seen any real use cases yet. Right now, OpenStack is more focused on how to better integrate with containers, Kubernetes, and helping telcos accelerate network functions virtualization delivery.