X
  • July 27, 2018

Here's what we learned from the data of 30 million smart meters

Oracle Utilities recently tested our Meter Data Management solution in a benchmark test designed to emulate the smart meter processes and real business scenarios of the largest utilities in the world.

The benchmark included data from 30 million smart meters, reporting nearly 100 reads per day and configured to replicate 60 million data channels.

The result?

In a five-hour window[1], Oracle Utilities Meter Data Management solution processed over 2.9 billion meter reads and generated billing determinants for 1.35 million customers. That equates to an impressive read-processing rate of over 642 million reads per hour.

 

Building a Benchmark … The Details Behind the Data.

For readers interested in looking at the details behind the data, here’s how we built the benchmark.

Our 30 million smart meter replication was configured for a 15-minute interval channel and a daily scalar channel.

The data were generated using PL/SQL to closely emulate production scenarios, and our team took great care to ensure that the transactional data was distributed to mimic real production data.

 

 Data profile for test “Utility”

Oracle Utilities Meter Data Management and Oracle Utilities Smart Grid Gateway were set up to run on a half-rack Oracle Exadata Database Machine X5-2 and a half-rack Oracle Exalogic X5-2. 

The test executed typical validation rules for meter readings, 10 validation rules were used in this test. 

The test produced and validated billing determinants, for 1.35 million accounts, deriving time of use values from interval data for 1.2 million accounts and summarizing register reads for another 150 thousand accounts. 

Using Oracle Exadata Hybrid Columnar Compression, the overall storage footprint was reduced by 50-60% as compared to traditional Advanced Compression.

Smart Flash Logging, Infiniband Network between all nodes and excellent DB storage IO help enabled extreme scalability.

Technical Configuration for Benchmark

 


[1] To scale IMD upload processing further, it is recommended to create additional OSB-MDB Clusters. When fully utilizing the half-rack of Exadata with additional OSB-MDB Clusters, the IMD load time is estimated to be approximately two hours.

 

Join the discussion

Comments ( 2 )
  • Akshay Jain Wednesday, August 1, 2018
    Sky high benchmark, commendable
  • Lesław Monday, August 6, 2018
    Are there some more details available?
Please enter your name.Please provide a valid email address.Please enter a comment.CAPTCHA challenge response provided was incorrect. Please try again.Captcha

Recent Content

Oracle

Integrated Cloud Applications & Platform Services