In the coming years, as massive amounts of data get created at the edge, entrepreneurs will look at the edge infrastructure and solutions with fresh perspectives for innovation as opposed to the current thinking that edge is just an extension of the core cloud. The architecture requirements won’t be Cloud vs. Edge; but it will also NOT be ‘Cloud Extended to the Edge’. In fact, Edge might need its own local cloud with mesh networks - 'Edge Cloud' that could be different than current central cloud architecture.
We at Service Ventures believe that the “Edge” is more than just a bunch of devices in consumer's hand or just lots of sensors on the factory floors. Edge has many different perceptions and architecture requirements depending on who is building the solutions and who is using it. Edge can be Consumer edge or Enterprise edge or Telco Edge or MSP edge. From a solution perspective, it may mean a lot of devices to some, lots of edge data centers with servers/storage/networking to some, lots of fiber connections, new wireless and mesh technologies to some and lots of distributed computing nodes to some.
Gartner has forecasted that, while 91% of data is created/managed within centrally designed data centers today, within the next few years, 75% of the data will be managed/analyzed at the edge. Based on this, some experts have gone to forecast that current and existing Edge computing technologies will drive adoption of use cases such as Autonomous Vehicles, AR/VR & Content Delivery, Predictive Maintenance, Connected Homes, vRAN in Telecommunications, Edge Internet Exchange (IX), Network Functions in Enterprise, Surveillance & Monitoring, Drones, Healthcare, Smart Cities, Voice & Video Recognition, Virtual Desktop & Video Conferencing for Productivity etc. In contrary, we think that critical ground zero uses cases will dictate how edge technologies gets built, adopted and evolve later over the coming years. Money/profit first, then the coolness of edge technologies and we might be just starting a long journey towards a pervasive edge computing paradigm.
From all the thought leadership that exist around edge computing today, we know that current central cloud based internet architecture will struggle to adapt to the needs (scalability, zero-touch, low cost, data intensiveness, latency, real time etc.) of the myriad edge computing use cases. Simply extending current central cloud-based technologies - just making them lighter and creating miniature versions of large central clouds will fail, precisely due to needs mentioned earlier. But we expect most cloud vendors and Hyperscalers to take this cloud-extended-to-edge path just to drum up the noise around edge computing, when their real intention is to siphon off as much as data as possible from edge to cloud. Part of this market confusion will also be driven by immediate economics of large cloud providers and Telcos. IBM recently purchased Red Hat primarily for a core cloud play - and IBM will not have any near term incentive to go after the edge from a fundamental value proposition standpoint, until they drive large amount of revenue out of the current acquisition through a core cloud architecture. Early edge computing customers will fall for this noise trap from Hyperscalers, assuming they are getting the purpose-built edge computing solutions they need, when they get are gimmicks of edge solutions. And lastly, VCs will keep investing in startups with such 'extension' technologies, not knowing the true run way of these architectures, because customers will eventually learn about the inefficiencies of such extension technologies from their use cases and will look for real edge computing solutions that are purpose built for edge use cases ground up.
What does this mean for us at Service Ventures to invest in edge computing startups? For starter, we believe that a new set of ecosystem players, other than large cloud providers and Telcos will emerge to honestly drive the edge computing market grounds-up with validated use cases in early adopting verticals. We will back edge computing startups that have focused solutions, right economics, and clearly defined playbooks for the ecosystem to scale fast. We forecast that the creation and consumption of massive amounts of data in IoT to create any sense of business value will need five enabling technologies — 1. Purpose built IoT HW and Devices, 2. 5G and other Communication technologies for Connectivity, 3. IT Infra for Edge Cloud and DCs, 4. AI/ML and other Data Analytics stack, and 5. Industrial Cyber Security.
We believe that only a unique mix of technology + economics will make it practical to assemble, store, and process data at the edge leading to the first set of immediate business insights leading to larger business value proposition and subsequent customer traction. Here are a few considerations we think for startups should keep in mind as they position their solutions for edge computing market:
Bandwidth & Latency: Typical edge environments will transmit all sorts of operational data and content data to edge servers/gateways before any cleaning and further optimization of date is carried out at the edge. Whether you are transmitting clean manipulated data from cloud to edge or collecting raw data from the edge and pushing to cloud, does not matter. So, the connectivity on the last mile has to be robust with low latency to offer any level of QoS.
No-touch & Self-healing: This is absolutely a critical feature when it comes to keeping the operational and ongoing cost of edge computing solutions low. Remote orchestration and operation from an Edge Cloud or DC will be needed to detect issues at the edge and heal without any sophisticated IT staff's help onsite. A bit of HW robustness and SW will be essential to handle such needs.
Scalability with Dynamic Resource provisioning: Architecture should have capability to dynamically provision IT infrastructure (compute/Storage/Networking) resources on-demand and distribute/split workloads across multiple edge locations as needed.
Harsh Operational Conditions: Unlike today's centralized data centers where there is continuous power, proper cooling, 24X7 connectivity, and on site staff to look after the IT infra, Edge architectures will need to be designed bottom-up to withstand extreme weathers, natural events and unmanned locations w/o guarantee of continuous power, cooling and connectivity.
Cyber & Physical Security: Starting from deploying the solution in a physically secure way to guaranteeing cyber security of end-to-end architecture and data (at rest, in motion), edge computing solutions must be built with security in mind for each layer from edge to the core. Specially if there are any Opensource components in the architecture.
Data Governance & Compliance: Although it seems that edge computing is a local issue with data stored on local nodes and acted upon locally, the data should always be treated as a rare global asset regardless of its origin. At some point down the line this data might be analyzed at a central cloud, by third parties to generate business value and hence GRC features needs to be in place upfront to manage the data privacy through its entire life cycle.
Comments