1. Cloud, Fog, and Edge Computing
5/4/2019
Cloud, Fog, and Edge Computing
CptS 464/564
April 24, 2019
Template I-Aqua curve

2. Sources of Info
cloudcomputing.ieee.org. Includes some training classes.
Kai Hwang, Geoffrey Fox, and Jack Dongerra. Distributed and Cloud Computing: From Parallel Processing to the Internet of Things, Morgan Kauffman, 2012, ISBN
T. Gamage et al. “Mission Critical Cloud Computing for Critical Infrastructures”, in Smart Grids: Clouds, Communications, Open Source, and Automation. D. Bakken and K. Inewski, eds, CRC Press, 2014, ISBN
Stuff available there includes:
“The NIST Definition of Cloud Computing”, Special Publication
“US Governmetn Cloud Computing Technology Roadmap Volume II: Useful Information for Cloud Adopters”, Special Publication
Wikipedia.org articles on “Cloud computing” and “MapReduce”

3. Sources of Info (2)
M. Ambrust et al, “Above the Clouds: A Berkeley View of Cloud Computing”, TR UCB/EECS , February 10,
Cloud Computing Tutorial,
“What is Cloud Computing – A Tutorial”, Short and sweet and to the point.

4. Over-Hyped, Under-Defined Terms
Cloud
Service-Oriented Architecture (SOA)
…. many more on the early side of the Hype Cycle!

5. Cloud Computing The “next new thing”
Big data centers (probably hosted by power industry vendors or NERC or DHS/DoE, not Amazon or Google)
These permit “consolidation”
10x or better reductions in cost of operation
Far better equipment utilization and management
New styles of elastic computing, potential to compute directly on massive data collections
Adds up to a new way of computing that forces us to undertake new kinds of thinking
But deliberately designed to trade off consistency for highest possible scalability

6. Clouds From a Hardware Point of View
Illusion of infinite computing resources available on demand
Can start small with no need to provision far ahead
Elimination of up-front commitment by Cloud users
Saves precious capital
Can start small: what I call “just-in-time expansion”
Ability to pay for use of computing resources on a short-term basis as needed

7. NIST’s Cloud Essential Characteristics
On-demand self-service
Broad network access
Resource pooling
Rapid elasticity
Measured service

8. IBM’s and a Book’s Cloud Definition
“A cloud is a pool of virtualized computer resources. A cloud can host a variety of different workloads, including batch-style backend jobs and interactive and user-facing applications.”
“Cloud computing applies a virtualized platform with elastic resources on demand by provisioning hardware, software, and data sets dynamically.” [emphasis mine]
Placement problem is huge here (DS II from [CDKB5] chap. 2)
Computer scientists would love to pass in info to influence placement and be told what mapping decisions were made

9. MapReduce
Programming model built for processing and generating large sets of data (e.g., Apache Hadoop)
Parallel & distributed algorithm on a cluster
Map() performs filtering and sorting
Divides problems into smaller sub-problems based on key & distributes them to worker nodes
Reduce(): Collects answers to sub-problems and combines in a given way for output
Much more scalable than sequential algorithms

10. Cloud Deployment Models
Private Cloud: for use by a single organization
RTE France has its own data center
Public Cloud: open for use by public, many companies and users using simultaneously
Community Cloud: operated for a specific community of consumers with shared concerns
E.g. ISO New England is starting one, will involve others
This is ultimately what grids need, IMO, managed right
Hybrid Cloud: mix of more than one of the above 3.
Often private us used, but then public for overflow demand

11. Service Models (Bottom Up)
Infrastructure as a Service (IaaS)
Lets consumers provision processing, storage, network, data, Operating System, etc. Low-level.
Platform as a Service (Paas)
Provides much higher-level APIs: langauges, libraries, middleware, etc. (c.f. “platform” from DS II section, from [CDKB5] chap. 2).
Client Apps
SaaS
PaaS
IaaS
Virtual Machines
Software as a Service (SaaS)
Provider’s apps running in cloud. Often accessed by a thin client on a client machine (web browser, simple app on phone, …)
Popular examples include Google’s gmail and docs, Microsoft Sharepoint, … $10B in 2010, $21B by 2015 (Gartner)

12. Service and Deployment Models
Saas
Paas
IaaS
Private Cloud
Apprenda, Stackato
VMware, Hyper-V, OpenStack, CloudStack
Public Cloud
Salesforce.com, QuickBooks online, Office 365
Google AppEngine, Microsoft Azure, Vmware, CloudFoundry.com
Amazon EC2, Rackspace
Community Cloud
Bakken:ElectricCloud
NYSE Capital Markets Community Platform
Hybrid Cloud
Custom CloudFoundry
Custom, Rackspace
(courtesy of Leverhawk.com)

13. Risks of Cloud Computing
Security & Privacy
Vendor Lock-In
Platform Lock-in: APIs not portable
Data Lock-in: your data is stuck with a vendor who may legally own it
Tools Lock-In: Tools to manage and develop cloud apps may not work across multiple vendors
Isolation Failure
Management Interface Compromise
Insecure or Incomplete Data Deletion

14. PANEL: Edge Computing for a More Adaptive and Manageable Grid
Chair: David Bakken, Washington State University
Paneilsts
Prof. David Bakken, WSU
Dr. Clifton Black, Southern Company
Mr. Byron Flynn, GE
Mr. Erik Felt, Real Time Innovations (formerly GE)
IEEE Innovative Smart Grid Technologies (ISGT) 2018 Panel
February 22, 2018

15. Fog Connects the Cloud to the Edge
Fog-enabled Router
Fog router CPU Coprocessor
Substation IED
Substation computer
IoT “thing”
with a Managed Infrastructure

16. Fog’s Raison D'être
Sources: Fog Computing and the Internet of Things: Extend the Cloud to Where the Things Are, Cisco Whitepaper, 2017; Nebbiolo TR

17. Fog Definition (cont.)
Fog Computing seamlessly extends cloud computing into edge for secure control and management of domain specific hardware, software, and standard compute, storage and network functions within the domain and enable secure rich data processing applications across the domain.

18. From Cloud to Fog to Edge
Source: Fog vs. Edge Computing, Nebbilio Technologies whitepaper, 2018.

19. Key Features: Fog and Edge
App Hosting
Yes
Limited
Data Services at Edge
Data & App Mgmt
Security
E2E, Data Protetion Sessisons & Hardware level
Partial point solution, VPN, FW
Elastic Compute/ Resource Pooling No
Modular Hardware
Virtualization with Windows support
TBD
Real-time control & high availability
Source: Fog vs. Edge Computing, Nebbilio Technologies whitepaper, 2018.

20. Differences: Fog vs. Traditional Edge
Edge Computing
Fog Computing
Awareness of devices and few services; unaware of the entire domain
Device independent, intelligent, and aware of entire fog domain
Limited control in the edge domain
Controls all devices in the domain
Cloud unaware
Extends cloud to Fog level as a continuum
Limited network scope
Complete network scope
No IoT vertical awareness
Supports and enabler for multiple IoT verticals
Uses Edge Controllers that are focused on edge device command and control
Uses Fog Nodes that are very versatile and capable of performing a variety of functions like RT Control, Application Hosting, and management
Security scope is limitied to devices
End-to-End (E2E) security
Analytics scoped to a single device
Fog Analytics enables collection, processing, and analysis of data from multiple devices in the edge for analysis, machine learning, anomaly detection and systems optimization.
Source: Fog vs. Edge Computing, Nebbilio Technologies whitepaper, 2018.