Keynote highlights from CEO Keshav Pingali’s talk at the Knowledge Graph Conference 2021, “High-Performance Knowledge Graph Computing on the Katana Graph Platform.”
In his keynote address at The Knowledge Graph Conference 2021 Keshav Pingali, CEO and co-founder of Katana Graph, discussed the key features of the Katana Graph Engine that enable high performance. He also discussed some important use cases for this technology from Katana's customers.
Keshav began his keynote by relating that, in a startup, regardless of how well-funded the company is, the leadership will always wear many hats. This includes the leadership at Katana Graph. In addition to being the CEO at Katana Graph, Keshav is a Computer Science Professor at the University of Texas, Austin.
In his keynote, Keshav described the reasons for graph computing, presented use cases, and discussed performance case studies.
Key take-aways:
- Unstructured data makes up 80% or more of enterprise data, and is growing at the rate of between 55% and 65% per year according to Datamation.
- Data professionals are losing 50% of productivity per week because of analytics complexity and data analysis time.
- Unstructured data can be viewed usefully as graphs of different kinds.
- Graphs consist of nodes that represent entities and edges that represent relationships between those nodes.
- High-performance graph computing provides the ability to handle higher volumes of data with faster times to insight.
- If you can analyze your data for insights immediately you will have a competitive advantage.
- Katana Graph uses high-performance graph computing to provide high-speed analysis of unstructured data sets.
- Katana Graph supports labeled property graphs, in which the labels have well-defined meanings.
- Katana Graph has customers that want to deal with graphs that have up to a trillion edges.
- Property graphs are useful to several industries including financial institutions, retail, telecom, industrial, energy, social networks, platform, and healthcare.
- Applications include fraud detection, anti-money laundering, customer 360, predictive monitoring, and precision medicine.
- Graphs have a role in mining for forbidden patterns in interaction graphs, identity governance and role mining, roles, and access control, electronic automation and circuit design, and medical knowledge graphs.
Lessons learned from use cases:
- Scale-out solutions are necessary for fintech, security, and identity, which require big graphs with lots of properties on nodes and edges.
- Single machine performance is essential for Exploratory Data Analysis (EDA) and pharma, in which graphs sizes are relatively small.
- Efficiency is critical when ingesting terabytes of data into a system can be a bottleneck.
- Querying and analytics should be tightly integrated.
- Solution must be cloud ready.
Learn more about how Katana Graph can help your business access massive amounts of data. Set up a meeting and demo with us.
