Behind the scenes at Labs
Showing results for 
Search instead for 
Did you mean: 

Watch the demo for Time-Varying Graph Analytics from Discover



By Curt Hopkins, Managing Editor, Hewlett Packard Labs

A big trend in big data is graph analytics. Graph analytics allows for an immediate, visual understanding of relationships within large, highly-connected data sets.

Telecommunications, security, online gaming, and ecommerce each create records in the trillions. Even if you’re only parsing a small percentage of the data in your industry, you’re handling more than ever.

A tool to handle these is time-varying graph analytics.

In this demo, led by Labs’ Omer Barkol and Sagi Schein, you can see how time-varying graph analytics will work in conjunction with The Machine. 

“A Time-Varying Graph is a way to model the data in a manner that will allow you to cope with querying of events where both the time aspect and the context are of interest,” said Labs research manager Omer Barkol.  “Modeling the data as a graph provides a way to capture context. Here a graph is a set of transactions between those entities.”

Analyzing a graph over time can provide important context to derive insights. Graph analytics can be likened to looking a photograph, whereas Time-Varying Graph analytics is like watching a video:

Imagine a photo of a golf ball in flight. From that static image, you have no idea where it came from, where it's going, or how fast it's moving. Now imagine a video of the same golf ball. Its origin, speed and destination become obvious.

Discover Insider, our tech conference publication, expressed The Machine advantage like this.

Most conventional computers can’t handle the massive computational power and intricate parallel processing algorithms required for graph analytics (but) The Machine’s massive memory and blazing fast fabric allows you to obtain a full history of data capturing at every state of the graph and across large chunks of time. Time-varying graph analytics will detect what’s normal, what’s changing, and what’s abnormal, and generate a graph database that is up to 100 times faster than existing technologies.

“Graphs are much better handled by in-memory systems as their traversal includes many random-access approaches,” added Barkol. “Thus, scaling up would be ideal but is limited in its volume. Most standard systems suffer from a lot of communication overhead for each graph traversal.

At Discover, Barkol and Schein showed an example of how the huge amounts of data that are collected in IT systems to perform security tasks can be viewed as a graph. The scales are enormous; the team queried this graph in order to investigate an event within the time-window and the context of that event.

“We showed how The Machine will be the right platform to do this type of work,” said Barkol.  

Watch the Time-Varying Graph Analytics demo here

0 Kudos
About the Author


Managing Editor, Hewlett Packard Labs

Online Expert Days - 2020
Visit this forum and get the schedules for online Expert Days where you can talk to HPE product experts, R&D and support team members and get answers...
Read more
HPE at 2020 Technology Events
Learn about the technology events where Hewlett Packard Enterprise will have a presence in 2020
Read more
View all