The astounding numbers that you can see on the previous page present a digital and IoT driven world that would require a new paradigm of datacenters to be able to meet the changing and growing demands. Today’s datacenter is being hurled as a centralized hoard of machines lined up on aisles and surrounded by huge walls, but going forward it will be architected differently to cater to the changing needs.
The future would be marked by the capability to ingest, process, and provide analytics on ZB of data that billions of devices connecting in and out of the network would generate every minute. That may mean taking the datacenter functions closer to the end user via edge architectures (ingesting, computing, processing and analyzing) combined with cloud or other technologies. Data-driven datacenter management architectures could be the ﬁrst step toward bringing about that change in the datacenter itself. While IoT remains the best use case for such a drastic makeover, intuitive/interactive customer experience, context-aware services, and associated applications would also embrace Edge-based architectures to have ‘lightening’ decision-making capabilities on datacenter networking. These certainly cannot be served with current datacenter models owing to latency, efﬁciency, and cost issues.
Considerable performance upgrades and efﬁciency are expected as data/insights move closer to the end point. Modular or micro-modular datacenters that can enable such distributed workloads or RoBo use cases will also gain more acceptance as users look forward to more real-time or near real-time insights or functions to serve their customers with an ‘instant gratiﬁcation’ kind of experience.
This will also mark a change from an integration of specialized or proprietary components that continuously demanded resources (human or technical) in large quantities in the previous models to a business-focused datacenter management architecture. The shift would essentially be from having mismatched components or solution stacks to an off-the-shelf, self-healing, easy-to-manage and AI/ML-based approach that may always not be kept/operated together.
While analytics and insights move closer to the end user, cognitive has already started to become an integral part of a datacenter. Whether traditional, Edge, or modular datacenter, the AI/ML component would bring in the two-fold ability to enable customized crunching and provide actionable insights as well as make the core datacenter itself self-healing, smart, and secure. These self-managing datacenters are expected to reduce human-machine interaction as well as datacenter operations downtimes. The next important change would be the advent of industrial clouds in datacenter hosting. Industrial clouds are in a nascent stage of development and concentrate on customizations according to a vertical industry (such as FSI and manufacturing) that takes into account speciﬁc legal, regulatory, and business and security requirements. The smallest building blocks of such datacenter models could be SDI, HCI, containers, GPU computing, or new forms of ﬂash storage (packed with higher densities). Such highly distributed, dense compute and analytics capacities would also entail a new form of energy and efﬁciency metrics (essentially beyond the green and PUE concept) and would require reliable, secure, and resilient power sources. That could mean the advent of micro-grids to support datacenter solutions such Edge or other architectures, signaling a change in the associated ecosystem as well. It is envisaged that there will be higher densities but greater efﬁciency in the future datacenter constructs along with a possible mix of energy sources (for example, solar) that may be used to meet these power demands. Lastly, all this would be based on the shift from a physical to a logical construct of a datacenter that would keep the functions together but not necessarily the components. This would further be enabled by a consumption-based, IT-driven model (utility services/as-a-service model) marking a tectonic shift in the datacenter ecosystem altogether.