[Din] revisit arguments for edge

[Aaron Ding <[e-mail]>]

for colleagues on lists, this 'Revisit Edge' is now published on IEEE Internet Computing:

https://ieeexplore.ieee.org/document/9470987

Aaron

________________________________
From: Sharon Barkai <[e-mail]<mailto:[e-mail]>>
Sent: Sunday, June 27, 2021 13:22
To: Hesham ElBakoury
Cc: Aaron Ding; [e-mail]<mailto:[e-mail]>
Subject: Re: [Coin] revisit arguments for edge

"The Automotive Edge Computing Consortium (AECC) estimates that more than 30% of video data produced on the vehicle will need to be offloaded. This is to increase safety thresholds by processing offloaded data with external data for augmenting awareness of the moving vehicle. The volume of data that will need to be offloaded is expected to be between 400 GB to 5 TB per hour. If all the data is sent to the cloud, the QoS objectives cannot not be met due to the limited bandwidth. Therefore, exploiting the edge that efficiently processes the data near the source is required for such applications.”

This “physics” argument is one of the most compelling for a programmable edge:
- not satisfied by CDN/rCDN edge caches with generic traffic management
- crowd-source collection of transient street conditions must be edge curated
- transient conditions such as blockages, free-parking, double-parking, hazards
- curation includes multiple source verification, localization of what is “seen”
- plus tracking for how long transient conditions persist and who needs to know
- all of which have policy based, changing and constantly evolving, enterprise logic
- attainable today on existing cellular with existing metro DCs peering to carriers

This programmable edge use-case is also very high-function-network COIN oriented:
- requires continues p2p p2mp service while modems switch carriers for reception
- requires addressable in-network instance steering to avoid resolution over mobile
- addressable vehicle IDs and addressable geolocation services while driving
- in-network instance selection, avoids massive client updates upon cluster shifts


On Jun 27, 2021, at 13:45, Hesham ElBakoury <[e-mail]<mailto:[e-mail]>> wrote:

The article answers this question in its conclusion which says: "This article argues that the motivation for edge computing research has not diminished since it was first formulated."

Hesham

On Sun, Jun 27, 2021, 1:55 AM Aaron Ding <[e-mail]<mailto:[e-mail]>> wrote:

COINRG colleagues,

Is the motive for Edge diminishing since its first concepts were formulated more than a decade ago?

A recent work to share on "Revisiting the Arguments for Edge Computing Research"
https://arxiv.org/pdf/2106.12224.pdf<https://urldefense.proofpoint.com/v2/url?u=https-3A__arxiv.org_pdf_2106.12224.pdf&d=DwMFaQ&c=XYzUhXBD2cD-CornpT4QE19xOJBbRy-TBPLK0X9U2o8&r=Iui-R0PdSAiRXYstq499vMjc_0AA9xFpAzYBY5yBmJc&m=hogSKtumATkFKILl92xzEruzN6Hei5e2FbdsI88owPk&s=TTTRoxwftZyC_6USOtaMuzmM_f57Ul4JVHepY4Qq6Ic&e=>

Cheers,
Aaron
--
CPI Lab, TU Delft
https://cpi-lab.github.io/<https://urldefense.proofpoint.com/v2/url?u=https-3A__cpi-2Dlab.github.io_&d=DwMFaQ&c=XYzUhXBD2cD-CornpT4QE19xOJBbRy-TBPLK0X9U2o8&r=Iui-R0PdSAiRXYstq499vMjc_0AA9xFpAzYBY5yBmJc&m=hogSKtumATkFKILl92xzEruzN6Hei5e2FbdsI88owPk&s=myyMyUBckLoqrPcp53dxhoQmZpn6z83EXzkoGI8eUwU&e=>