[oe-sync] Latency + need for edge computing

[Klas, Guenter, Vodafone Group]

The ITU report from 2014 (on tactile Internet) is also a good reference that firmly suggests the need for edge computing with response times down to 1 ms for certain application areas. Though mostly written by a German research community, I think many statements underpin the need for very low latency.
https://www.itu.int/dms_pub/itu-t/oth/23/01/T23010000230001PDFE.pdf 

Guenter

-----Original Message-----
From: oe-sync-bounces+guenter.klas=vodafone.com at lists.andrew.cmu.edu [mailto:oe-sync-bounces+guenter.klas=vodafone.com at lists.andrew.cmu.edu] On Behalf Of Padmanabhan Pillai
Sent: 04 September 2015 14:22
To: oe-sync at lists.andrew.cmu.edu
Subject: [oe-sync] Latency + need for edge computing

In the last sync meeting, the question of latency and showing the need for edge computing (as opposed to cloud computing) came up. The CMU team has done extensive work already on demonstrating the need for edge computing, and have a published paper on this topic (see attached).  The work looks at a few interactive, compute-intensive applications, and compares mobile-only execution, offload to several different Amazon datacenters, and offload to a local edge server.  In short, the paper shows that generally, edge computing comes out ahead in these interactive application scenarios.  I hope you can use this work as supporting evidence when encountering naysayers questioning the need for edge computing.

The paper was presented at the International Conference on Cloud Engineering (IC2E), 2013.

Regards,

- Babu