Explosive growth of Smart Meter (SM) deployments has presented key infrastructure challenges across the utility industry. The huge volumes of smart meter data has led the industry to a tipping point which requires investments in modernizing existing data warehouses. Typical modernization efforts lead to huge capital expenditures for DW appliances and storage. Sizing this new infrastructure is tricky and can lead to underutilized or poorly performing hardware.
The Cloud is the catalyst to solving these Big Data challenges.
Utilizing a Cloud architecture delivers huge benefits by:
Maximizing use of existing architecture
Minimizing new CapEx expenditures
Lowering overall storage costs
Enabling scale on demand
Supercharging Smart Meter BIG DATA Analytics with Microsoft Azure Cloud- SRP Case Study
1. Session 403
Supercharging Smart Meter Big Data
Analytics with Microsoft Azure
Jason Wilhite, Manager, Enterprise BI Services, SRP
Kirk Nason, Director, BI Solutions, Neudesic
2. Salt River Project (SRP)
SRP is one of the nation's largest public power
utilities. Providing electricity to more than
984,000 retail customers across three Arizona
counties
SRP is an integrated utility, providing
generation, transmission and distribution
services, as well as metering and billing services.
SRP's water business is one of the largest raw-
water suppliers in Arizona. We deliver about
800,000 acre-feet of water annually to a 375-
square-mile service area and manage a 13,000-
square-mile watershed
3. Neudesic in the Utility Market
History & Experience
• Half a decade in the domain space
• 10+ dedicated technologists developing
solution & IP
Solutions & Intellectual Property
• Smart Meter Analytics Solution:
Big Data Hybrid Analytics Solution for the Utility
Industry
• Neudesic BI for SAP:
Integrates SAP data into the Microsoft Data Platform
• Neudesic BI BIG Data Framework
4. SRP’s Goal & Business Challenges
• Massive amounts of data
• Disparate data sources
• Market alternatives are cost
prohibitive
Maximize grid reliability, power delivery &
service through Big Data analytics
Rising Infrastructure Challenges
• Must solve the infrastructure
challenge to attack the business
problems
5. Smart Meter Analytics Solution
• Big Data Business Intelligence solution
• Processes data collected from
Smart Meters
• Built on the Microsoft data platform
• Built using Neudesic BI framework
• Augmented by Neudesic BI for SAP
Visualization
Domain
Model
ETL
Connectivity
Meter Data and other Line of
Business Data Sources
6. V1 Smart Meter Arch at SRP
850,000+ Residential Meters
• 15-minute interval reads
• 96 reads per customer per day
• 81.6M meter records added per day
• 2.6B rows a month = 160GB data
• Reprocess 30 day rolling window
nightly
New IT challenge:
ETL processing Service
Level Agreement (SLA)
began to slip
Smart Meters Meter Data
Management System
ETL integration
BI Reporting & Analytics
Customer data (CRM) MDMs Data
Store
BI Staging
Reporting and Analytics
SQL Server BI
Input files
7. V1 On Premises Data Prep SLA
6:00 PM 8:45 AM
7:00:00 PM 8:00:00 PM 9:00:00 PM 10:00:00 PM 11:00:00 PM 12:00:00 AM 1:00:00 AM 2:00:00 AM 3:00:00 AM 4:00:00 AM 5:00:00 AM 6:00:00 AM 7:00:00 AM 8:00:00 AM
6:00 PM
MDMS Load Start
1:59 AM - 3:47 AM
Validation (1:48)
9:33 PM - 1:43 AM
Derived Facts/file extract (4:10)
1:43 AM - 1:59 AM
Cube Processing (00:16)
6:00 PM - 9:33 PM
Staging and Base Facts (3:33)
8:00 AM
SLA Availability
SLA Window is 6pm to 8am = 14 hours
Smart Meter Fact Processing
• Staging & prep base facts
• 81.6M meter records added daily
• Reprocess 30 day rolling window nightly
• 2.6B rows a month = 160GB data
• Create derived gap facts
• 188K rows daily = 20MB
• Load new facts into Analysis Services
Derived facts are now taking 6 to 12 hours
8. Requirements to address
the SLA Challenge
Maximize use of existing architecture
Minimize new CapEx expenditures
Enable Scale on Demand
Lowering overall storage costs
Protect Personally Identifiable Information
9. V2 Addressing the SLA Challenge
Azure Hybrid Smart
Meter Solution
Smart Meters Meter Data
Management System
Base Facts and
Confidential data BI Reporting & Analytics
Reporting and Analytics
SQL Server BI
Customer data (CRM) MDMs Data
Store
BI Staging
Blob Storage
HDInsightInput: Base Facts
Extract meter reads from data
mart and upload to blob
storage
Download derived
facts
Output: Derived Facts
Input files
2.6B Rows
160GB
188K Rows
20MB
81.6M Rows
3.5GB
10. Azure HDInsight Catalyst for Success
Hyper Scale on demand
Example:
• 32-Node cluster running 1hr/day
• 15TB Storage & Data movement
• ($357.12+$354.41)=$700/month
($8,400/year)
Lowest Cost of Storage
11. Smart Meter Analytic Solution Content
Solution Contents:
• SQL Server EDW Utility Schema
• SSAS OLAP Cube Utility Schema
• SSIS ETL Packages
• Azure HDInsight: Hive Queries & Scripts
• Neudesic BI Framework
• SharePoint Server BI Portal
• Excel Reports
• Implementation Services
12. Q&A
Contact Info:
Kirk Nason
Director, BI Solutions
Kirk.Nason@Neudesic.com
(949)789-2683
Jason Wilhite
Manager, Enterprise BI Services
Jason.Wilhite@srpnet.com
(602) 236-5528
Jason to cover. SRP overview slide, feel free to tweak.
Jason to cover - Take up to 5 minutes
Goal around introducing Smart Meters data into SRP. Initial thoughts on getting to v1.
Talk about initial solution architecture and how Big Data let to new infrastructure challenges.
Frameworks:
Neudesic Large Structured Data Load (Partition-switching)
SSAS Cube Partition Mgmt & processing
SSIS ETL Framework
Dynamic OLAP Security
Talk to Orion about the SLA and why this box is taxed..
PII = personally identifiable information
Maximize use of existing architecture
Minimize CapEx
Enable Scale on Demand
Lower storage costs
Protect PII data
Maximize use of existing architecture
Minimize CapEx
Enable Scale on Demand
Lower storage costs
Protect PII data