Modern Architecture for the Modern Smart Grid

chirpskulkInternet και Εφαρμογές Web

3 Νοε 2013 (πριν από 4 χρόνια και 6 μήνες)

93 εμφανίσεις


Architecture for

the Modern Smart Grid

Brad Molander

Technology Evangelist @NISC

Software Engineer for 15 years

9 years at NISC

Various Projects:

Billing, Mobile Phones, Crew Management,

NISC Social Community, Document Imaging

More Recently:

Grid Computing in the Private Cloud

Meter Data Management / Smart Grid Solutions

Native Mobile Application Development

Apply technology to everyday challenges

Understanding Cloud Computing

Overused term, ambiguous

Several different types, let’s explore

World is


World is


People do not delete data

They expect to keep it indefinitely

Need to understand their data

Search, analyze, categorize

People are mobile, so are their

By 2013, majority of Internet traffic will
be from mobile devices

iPad, iPhone,

Keep Big Data available to people
who use it

Cloud Provider Types



Securing Data

Who owns the machines?

With whom are you co

Determine your property visibility up front
in your SLA



Solid solution for backups

Hardware operations

Resource sharing

Traditional Software


Does not vertically scale well

Always limited by physical server constraints

Hosted Services

Host a server at a co



High performance

Distributed Computing

No Virtualization

Dozens or thousands of

machines working together

as a single unit

Commodity servers

Private Cloud Types

Cloud Offerings


(Software as a Service)

, Microsoft Online Services,

Ready to use applications


(Platform as a Service)

Google App Engine, Azure,

Databases, number

crunching, application



(Infrastructure as a Service)

Amazon Elastic Cloud, Rackspace, Cooperative Cloud

On demand hardware, clusters of hardware

Public Cloud Types

Smart Metering by the Numbers

100,000 Meters

2.4 Million measurements a day

72 Million measurements a month

876 Million measurements a year

8.7 Billion measurements over 10 years

Who wants the data?

Engineering Analysis

Operational Audits

Consumer Presentation and Analytics

Demand Response

Load Forecasting

Why Distributed Computing?

Vertically Scaling is costly and limited

Big Data is constrained by Disk IO and Memory

Increasing Performance




Mining Data

20 years ago, keeping this data was impossible

10 years ago, we used lots of

Compact Discs

5 years ago…

What if
we could

analyze 1 billion

rows in
10 minutes

Commodity Hardware

Instead of vertically scaling servers, find
the sweet spot for pricing

Keep them fairly identical



Data Center One

Data Center Two

Grid Computing

Ability to spread tasks
and data across
thousands of nodes

Run computing
network just like an
electric grid

Similar to Multi
Core Revolution

When multi
core CPUs shipped
several years ago, only a few
applications took advantage of

Your applications have to be
written from the the ground up to
support cloud architecture

Breaking Down Tasks

100 Minutes

1 Server

Breaking Down Tasks

5 Minutes

20 Nodes


More nodes yields:

ormalized distribution of load

Better resiliency

Higher fault tolerance

Net Effect

We can do things that we could not do before

Monitoring the Grid

Managing the Grid

Internal Operations

Machines can and do fail

They can be replaced on the fly

Little to no impact on cloud services

Effects of Elasticity

Support Big Data applications


Electrical Analysis

Consumer presentation

Also support traditional applications

New Possibilities

Engineering Analysis

Indefinitely store analytics

More accurate predictions based on

Nationwide energy profiling by

Now What?

Nearly all applications will
be a natural fit for a
Hosted Cloud

Ensure network continuity
plans, have a backup for
your backup

Will Cloud work

for you?

Identify business processes

Train employees on best
practices for network

Have backup plans for trivial
business processes