Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Dogs Never Get Tired: Power and Edge Computing
Usenix Annual Technical Conference (ATC) 2021Session Chairs: Marcelo Martins, Apple
Dilma da Silva, Texas A&M UniversityPreview by Ketan Bhardwaj, Georgia Tech
Once there were a few big dogs …
Had to haul more, Were a bit slow …
IoT
The big ones already knew it …
Akamai CDN Netflix Open Connect Google Global Cache
Pushed content cloud closer to users• Reduce backhaul demand• Several 10s to few 100s ms• O(few 1000s) global locations
How fast do they really needed to be …
26Telefónica Alpha / Edge Computing /
Challenge #2: How many Edges?
1-20 20-80milliseconds milliseconds
1
Visual-tactile
feedback
m
s Human Features
m
s
Network &
Spatial
m
s
Use Case
Requirements
1
1
Tactile
Internet
5G (target)
1
Car at 140 km/h,
3.9 cm
Wearable
Assistance
4
Tell that a
sound is a human
voice
7
Vestibulo-ocular
reflex
Visual reaction
time
1
Speed of light
per 200 km
5
Augmented
Reality (gaming)
7
Autonomous
Driving
5
Two people in a room,
latency added by speed
of sound
8
Car at 90 km/h,
20 cm
TV max inter
picture
Virtual Reality
2
0
WiFi (average)
2
0
An airbag to
release
3
0
RTT UK N-S (IP)
(564 miles)
3
0
Perception of a screen
change after clicking a
button
5
0
4G/LTE
(average)7
0
RTT USA W-E (IP)
(2514 miles)
8
0
Screen to brain
propagation
10
0
Auditory reaction
time
13
0
3G (average)
15
0
Human blink
15
0
Crisp UI with a
computer
8
0
Augmented
Reality
(non gaming)
10
0
Tele
Robotics
Telefónica Alpha / Edge Computing / July 2017
1
0
1
0
1
0
1
0
80-150milliseconds
>150milliseconds
THE ADVENT OF EDGE COMPUTING
Many edges, per use case and workload
Source: Pablo Rodriguez, Telefonica Alpha, SEC’17
So, there were more but smaller ones … 1/31/20, 8)32 AMVerizon partners with AWS to bring more power to its 5G edge | FierceWireless
Page 1 of 3https://www.fiercewireless.com/5g/verizon-partners-aws-to-bring-more-power-to-its-5g-edge
Next Gen Wireless Summit Video Wireline Resources Webinars Events Subscribe
WIRELESS TECH 5G IOT
Verizon is bringing cloud computing closer to its 5G network edge by teaming with Amazon WebServices (AWS). Verizon will use AWS Wavelength, Amazon’s new cloud platform designed foredge computing, to make it easier for developers to create and deploy low latency applications.
Speaking at the Wells Fargo TMT investor conference, Adam Koeppe, senior vice president oftechnology and planning at Verizon, explained that Verizon is integrating AWS Wavelength atstrategic locations within its 5G network that uses the its 28 MHz mmWave spectrum. Wavelengthwill basically open up Verizon’s network to AWS developers that want to create low-latencyapplications that will run on a 5G distributed network. Examples of these types of low-latencyapplications include smart cars and cities, augmented and virtual reality, Internet of Things andgaming.
Koeppe said that customers are already testing Wavelength technology in Chicago where Verizonhas deployed its 5G network. The two companies said they plan to make the 5G edge computing
5G
Verizon partners with AWS to bring more power to its 5Gedgeby Sue Marek | Dec 3, 2019 4:33pm
Verizon CEO Hans Vestberg made an appearance at AWS re:Invent conference in Las Vegas to talk about how Verizon plans to incorporateAWS cloud compute and storage at the network edge. (AWS)
GET THE NEWSLETTERSubscribe to FierceWireless to get
wireless industry news and updatesdelivered to your inbox.
SIGN UP
I acknowledge that I may receive emailsfrom FierceWireless and on behalf of theirtrusted partners.
␣
" #$
1/31/20, 8)35 AMAT&T and Microsoft announce early result of their 5G-cloud partnership - CNN
Page 1 of 4https://www.cnn.com/2019/11/26/tech/att-microsoft-azure-5g-cloud/index.html
AT&T and Microsoft announce earlyresult of their 5G-cloud partnership
Updated 5:47 PM ET, Tue November 26, 2019
By Clare Duffy, CNN Business
New York (CNN Business) — AT&T and Microsoft have announced the next phase of a partnership that's aimed atadvancing development in two of the buzziest areas of tech: 5G and cloud computing.
The companies inked an agreement in July to work together on innovations in 5G and cloud. 5G is next generationnetwork technology that's expected to change the way people live and work. And the cloud is distributed datastorage and computing power that doesn't rely on traditional on-site data centers. AT&T (T) (CNN's parentcompany) also agreed to migrate its data and workflows to the Microsoft Azure cloud, in what was reportedly amore than $2 billion deal for Microsoft (MSFT), according to Reuters.
On Tuesday, Microsoft and AT&T's business unit announced the pilot launch of their so-called "network edgecompute" technology, an early outcome of the partnership. The advancement is expected to help both companies'enterprise customers take advantage of the benefits of 5G, as AT&T works to roll out 5G across the US by mid-2020.
"We've said all year, developers and businesses will be the early 5G adopters, and this puts both at the forefront ofthis revolution," AT&T Business Chief Marketing OWcer Mo Katibeh said in a statement Tuesday.
AT&T will implement the Microsoft technology in some of the major metro areas where it's installed 5G — AT&Tsays it will be available for select customers in Dallas early 2020, with Los Angeles and Atlanta slated forimplementation later in the year.
Microsoft and AT&T are excited about the technology's potential, but Gartner analyst Bill Menezes says it's not yetclear just how useful this technology will be and when it will be usable.
"Right now, it's creating a lot of noise that plays on a lot of current buzzwords in tech," Menezes said. "What itactually produces remains to be seen."
What is network edge compute?"Network edge compute" is a mouthful. Here's what it means:
5G promises to bring super fast download speeds, the ability to connect millions of devices and the ability toprocess massive amounts of data in nearly real time. But if those devices are connected to the cloud, reaping the
LIVE TV
1/31/20, 8)39 AMDeutsche Telekom Embarks on Edge Computing for IoT - SDxCentral
Page 1 of 3https://www.sdxcentral.com/articles/news/deutsche-telekom-embarks-on-edge-computing-for-iot/2019/12/
News
Deutsche Telekom Embarkson Edge Computing for IoT
Matt Kapko | Editor
December 3, 2019 12:22 PM
Share this article:
T-Systems, a subsidiary of global operator Deutsche Telekom, unveiled an edge
computing platform that delivers low latency for IoT applications in enterprise
facilities. The EdgAir platform is designed to operate like a private cloud to ensure
security and transmit data in less than 10 milliseconds, according to the IT services
and consulting company.
The platform is based on OpenStack and allows enterprises to run IoT applications as
virtual machines (VM) or in Docker containers as microservices that operate
independent of the underlying infrastructure. The containers are also coordinated via
Kubernetes and EdgAir equipment can be installed as a pole mount, ruggedized rack,
or a standard rack.
“In combination with Time Sensitive Networks and a very good service level
agreement, our managed platform EdgAir can meet requirements such as real-time
control of machines in a production hall,” said Albert Kroisleitner, senior product
manager of edge platform at T-Systems, in a prepared statement.
Learn More
Related Resources
Assuring Intelligence for the Multi-Vendor SD-WAN Edge: Webinar
Webinar: Assuring Intelligence for the Multi-Vendor SD-WAN Edge
Assuring Intelligence for the Multi-Vendor SD-WAN EdgeSponsored by Infovista
Subscribe to our Newsletter
Email Address
SUBSCRIBE NOW
Learn More
Sponsored Hubs Cisco Dell Technologies Versa Networks Get SDxCentral'sDaily Newsletter
SD-WAN SDN NFV CONTAINERS 5G SECURITY EDGE MULTI-CLOUD OPEN SOURCE IOT AI MORE +LOGIN
Where are these small ones …
Where are the edges?Distributed cloud, edge compute, AI/ML, IoT, 5G, VNFs/NFV, FMC
Source: IHS Markit. NFV Strategies: Global Service Provider Survey, June 2017; Respondents control 61% of global telecom capex
uCPE
Data Center Near CO
[Regional DC]
uCPE
Smart COwith Mini
DC/CDN+[Edge DC]
Data CenterNot Near CO[Central DC]
EDGE 97% of operators plan VNF
execution in Smart CO
PARTIAL EDGE
85% of operators plan VNF execution
in DC Near CO
NOT EDGE
70% of operators plan VNF execution in
DC Not Near CO
EDGE 82% of operators plan VNF execution on uCPE
at customer
sites
EDGE MEC server,
AI/ML, IoT, 5GEDGE
VNFs, vEPC, MEC, distributed
RAN, vRAN, BBU hotel, FMC, vCPE, AI/ML, IoT
go here
IoT G/W
6
EDGE
Enterprise & IoT
20msecs
IoT G/W
How do the small ones help …
What help do small ones need …
• Reduce latency to few ms• Save backhaul bandwidth
• Handle multiple applications• Power/energy constraints
This session
Prediction-Based Power Oversubscription in Cloud Platforms• Power is important • Power oversubscription is
limited • Prior work assumes hard
to get info • Reactive, expensive
mechanisms
• Prediction-based approach• Predict VM criticality &
utilization• Protect critical VMs by
intelligent placement • 2x oversubscription
Proactive Energy-Aware Adaptive Video Streaming on Mobile Devices
• ABR widely used video 360 streaming – App adaptation• Prior work, mostly reactive causes
adaptation oscillations• Difficult to predict energy usage• Energy use is cummulative
• Integrated, Pro-active ABR adaptations for video• Reactive vs. Proactive• Function wise power prediction • Practical control algorithm
Video Analytics with Zero-streaming Cameras
• A lot of cameras (surveillance)• Huge amounts cold video
(nothing interesting)• Especially for ad-hoc queries:
retrieve, tag, count• Why stream … even to edge• Cloud processing only needed if
more accuracy is needed
• Zero-streaming• Leverage spatial skew in
landmarks• Cloud provides increasingly
complex / more accurate operators based on policy• Multi-pass, multi-operators• 100x more video in realtime
ASAP: Fast Mobile Application Switch via Adaptive Prepaging
• Everyone uses more than one app • Switching b/w them can be faster
(useful)• Cause – slow demand paging• Requires careful low level design
to avoid overheads
• Predict likely to be accessed pages / pre-page• Ensure reasonable overheads• 20-30% faster switch • Additional 30% CPU & 25% disk
bandwidth usage
Enjoy the session J