FIGURE 1: Edge compute is a shift from a centralized compute model to a distributed compute model.
October/November/December 2020 I 27
“The next wave of technology innovation is already
here with new applications transforming the way we
live, work, and travel,” TIA says in its paper. “The huge
adoption of these new services drives exponential growth
in the total demand for data. We must provide more data
capacity and higher computing speeds if we hope to keep
up. The sheer scale and scope of the gap we face demands
that we rethink the way we have traditionally organized
the design and deployment of networks and data centers.
As many hands make light work, deploying many smaller
distributed data centers seems the most viable solution.”
These smaller distributed data centers, which complement
the cloud, are what the data center industry
today recognizes as the edge.
Edge compute pushes computing functions closer
to the end use, proximate to where data is generated,
analyzed, and stored. This evolution of the network realizes
numerous benefits, chief among them are reduced
latency between the end use and the compute resource
(cloud or servers), improved application performance,
and optimized transmission costs.
Applications for the edge are far-reaching. The
fastest growth is happening in applications that are most
sensitive to network latency, require enhanced security
or are bound by regulatory requirements. Industries such
as manufacturing, healthcare, and finance are among
those that are adopting edge networks today.
Regardless of the application or the reason for
the transition to the edge, there are ways to leverage
infrastructure that impact the deployment, monitoring,
management, and operation of an edge data center
(EDC). These infrastructure choices can help those
responsible for the data center to address common
challenges related to security, environmental conditions,
and reliability, thereby ensuring uptime for
the network and those systems that are dependent
on the network.
EDGE COMPUTE EXPLAINED
At its core, edge compute is a shift from a centralized
compute model to a hybrid distributed compute model.
What does that mean? For the past 20 years or so, the
compute model has been almost strictly centralized.
On-premise data centers, cloud compute, and hybrid
cloud models are all centralized compute models, meaning
all the data is processed and stored in a central location
(e.g., data center, cloud data center). Edge compute
supplements that model with additional compute
resources at the edge or in closer proximity to the
end use (Figure 1).
Centralized Compute Architecture Distributed Compute Architecture