To support 5G (fifth generation of wireless communications technologies),
experts and analysts suggest that the structure of the data center needs to change;
including the existing functions of routers and switches, whiles technologies
like NFV (Network function virtualization) and SDN (Software-defined
networking) must be introduced on
a larger scale - pointing strongly to the adoption of edge computing. Gartner predicts that edge computing will account for 75% of
enterprise-generated data by 2025 from 10% in 2018 [1]. Ericsson
predicts that 25% of all 5G use cases will be enabled by Edge computing [2].
What is
at the Edge of Edge computing
The number of
applications expected to run on 5G is astonishing, as more and more 5G business
cases are being explored and tested. As with most 5G use cases, huge volumes of
data are expected to be processed at the edge for efficiency and cost savings. For
instance, a Telco, based on analytics, may provide customized content based on the
location of university campuses, shopping malls, offices and homes. Home
locations may desire capacity intensive content like Netflix, whereas office
locations may require less-intensive services like email, news and social media
content. Uber or Tesla may have their own distributed edge data centers to
track their fleet of cars or to support autonomous driving. As we move towards this
highly geographically distributed mini data center ecosystem, where computing
is done closer to the user/data source (aka Edge Computing), a few requirements
come to mind.
- Collocation
at the Edge: The concept of collocation in Edge
computing is still relatively greenfield just as is Edge computing. It means
resource sharing down to the virtual machine level or hardware level-server/storage
or cabinet/pole/container sharing. Backing this with regulation will ensure a
well-controlled, nuisance-free ecosystem.
- Connectivity
for the Edge: 5G connectivity promises to offer Edge
data centers with the needed low latency, high throughput and high-speed
connections to the central location (core backbone) and to the clients (last
mile). For instance, a vehicle tracking application that could only collect
data from 100 sensors (100 cars) every 30 seconds, can now collect real-time
data from 1000 even 10,000 sensors every second.
- Power for the Edge: Edge data centers are projected to account for 102GW of power by 2028 [3] -a huge figure that poses a lot of concern for the success of Edge computing and 5G. A number of solutions come to mind:a. With cooling contributing about 38% to the total power consumption of a data center [4], energy efficient mini data centers that use technologies like Free Cooling may be deployed to save energy.b. Explore more efficient server/compute allocation strategies that ensures that not all servers are running at the same time-compute optimization.c. Collocation of resources/applications must be encouraged to lower power consumption.
- Management
for the Edge: Think about the effort and cost in
maintaining a single data center today. Now multiply that by 100, maybe 1000,
even perhaps 10,000 geographically dispersed mini data centers. Regular
maintenance means frequent trips to 1000 mini data centers. This brings to
light the need to automate maintenance at the Edge which promises to
drastically reduce human intervention that would ultimately trickle down to
actual cost savings. Adding self-diagnostic and self-healing capabilities into
the fabric of edge data centers will drastically reduce the associated costs
with maintenance and management.
Ensuring
security for the Edge
I am almost
certain that 99.9% of service providers will consider physical security (high
level surveillance, anti-tampering, alarming, etc.) of their mini data centers
as top priority. How about cybersecurity?
The promise
of 10 times faster data connectivity with 10 times lower latencies over the
internet, makes cybersecurity a critical factor in the success of 5G and edge
computing.
The worldwide
adoption of 5G technology, which will be predominantly wireless broadband will
mean that corporate networks that will rely on this technology need to adopt a
security technology that is not MPLS (Multiprotocol Label Switching) as we know
it. This is where SD-WAN (Software-Defined Wide-Area Network) comes in. Think
of SD-WAN as connecting your very important "gold" resources - this
time not through a private connection like MPLS, but through the public
internet, and still providing high level security for those connections.
Deploying
SD-WAN over 5G connections for enterprises is an inevitable reality. Orange and
Nokia have already demonstrated this capability [5].
Concluding
remarks
When you
think of the 5G/data center evolution, think Edge computing and how it can
enable your business propositions. And whiles discussing those business
propositions, think about security. Why have costly private WAN connections for
thousands and millions of Edge data centers whiles 5G offers better throughput
speeds and latency. The only problem is that it will be over the internet. This
is however solved with SD-WAN technology which is already here. Think of Edge
computing as the enabler for highly secured, highly resilient, and low latency
business cases and explore the possibilities that these technologies (Edge, 5G
and SD-WAN, together with Artificial Intelligence and other concepts) bring in
unison to create the perfect business model for today’s marketplace.
The
possibilities are enormous; however commercial viability of the business cases
must be quickly validated to understand the ecosystem better. With Telcos and
hyperscale cloud providers leading the charge, we can harness the full
potential of the various propositions in this promising era of 5G, Edge
computing and SD-WAN.
Author:
Kwadwo Akomea-Agyin; | Digital Solutions
Expert & Business Analyst | Member, Institute
of ICT Professionals, Ghana.
For comments, contact kojo.e@live.com
References
[1] Van der
Meulen, R. (2018, October 3). What Edge Computing Means for Infrastructure and
Operations Leaders. https://www.gartner.com/smarterwithgartner/what-edge-computing-means-for-infrastructure-and-operations-leaders/
[2] Beyond
edge computing with distributed cloud. (2020, April 6). https://www.ericsson.com/en/digital-services/trending/edge-computing
[3]
Fulton, S. (2020, March 24). Just How Much Energy Will Edge Data Centers
Consume? https://www.datacenterknowledge.com/edge-computing/just-how-much-energy-will-edge-data-centers-consume?NL=DCK-01&Issue=DCK-01_20200325_DCK-01_213&sfvc4enews=42&cl=article_1_b&utm_rid=CPNET000018484870&utm_campaign=24990&utm_medium=email&elq2=4109831742bb4efc87f6da6d332e57ef
[4] Mangu, A.
(2018, December 16). Managing Energy Consumption of Data Centers. http://large.stanford.edu/courses/2018/ph240/mangu2/
[5] Morales,
F. (2019, August 5). Orange and Nokia set a world first for SD-WAN over 5G. https://www.orange-business.com/en/blogs/orange-and-nokia-set-world-first-sd-wan-over-5g
No comments:
Post a Comment