10 things you should know about Multi-access Edge Computing (MEC) - Tech | 5G, SDN/NFV & Edge Compute


Monday, February 4, 2019

10 things you should know about Multi-access Edge Computing (MEC)

What is MEC?

MEC stands for Multi-access Edge Computing. It was originally called as Mobile Edge Compute. However, due to the general applicability of an edge compute infrastructure for additional wireline use-cases, it was later renamed as Multi-Access Edge Computing. MEC helps service providers to bring application oriented capabilities closer to the users and support several latency sensitive use cases from the edge. MEC plays a huge role in 5G networks, as 5G is expected to support less than 10 milliseconds latency. (Read: Eight 5G Currencies)

What are the benefits of MEC?
  • Reduces latency of applications 
  • Allows local policy management 
  • Provides security for IOT services, by distributing the attack surface
  • Improves reliability of application and network services, by offering a distributed infrastructure for service failover
  • Provides real time access to data locally, in an IOT environment
  • Reduces operational costs, by avoiding the need to build costly data centers
  • Improved application performance and user experiences, as applications are run in local compute infrastructure 
What role does MEC play in reducing the latency?

IHS Markit, a global research organisation, defines edge compute as being within 20 milliseconds of the end user or device. MEC is able to achieve such low latencies, by hosting applications and services near to the end user / device. 

Verizon recently announced that they were able to reduce application latency by half, by deploying MEC infrastructure. 

How does MEC help Service Providers to differentiate themselves over Over the Top (OTT) vendors?

Telcos are at a great advantage of supporting low latencies with their infrastructure presence in central offices and cell towers, when compared to OTT service providers. 

OTT service providers have partial coverage for edge compute with distributed data centers that are within the 20 milliseconds to 50 milliseconds range, while telcos can hit 5 milliseconds to 20 milliseconds.

A few service providers with an "engineering DNA" would build a home grown infrastructure for MEC. However, many service providers would end up deploying the MEC infrastructure from off-the-shelf MEC vendor products or by partnering with existing Cloud Service Providers. 

What are some of the use cases supported by MEC?

- Vehicle to Everything (V2X)  (Autonomous Cars)
- Radio Access Network (RAN) aware Video Optimization
- Virtual Reality / Augmented Reality
- Hosting Enterprise applications
- Edge Video Orchestration
- Massive IOT
- Surveillance & Security Services

How does MEC benefit autonomous cars?

If a network has a 100 milliseconds delay and a car is running at 120 km/hr, if the car detects danger, it would travel over 3 meters further before taking any action. This would potentially mean an accident or a car crash. In order to avoid such crashes or accidents with the current network, the cars need to have larger gaps. With 5G,  its single digit millisecond latency will make self driving scenarios safer, while improving road throughput (number of cars/vehicles handled by the road) at the same time.

Is MEC specific to 5G deployments?

No. MEC can also be deployed along with 4G networks. In fact, MEC can also be deployed in wired environments (for ex., within an enterprise) (Read: 5G Deployments Status)

What are some communities trying to standardize the MEC stack?

Open Edge Computing Initiative, Cloud Native Compute Foundation and ETSI are communities that are involved in standardizing the MEC stack. 

Who are the vendors creating an edge computing stack?

What are the open source communities for edge computing stack?

No comments:

Post a Comment