Bob Muglia is something of a database bard, capable of unfolding vast stories in the evolution of technology.
This is Muglia, former CEO of Microsoft and former CEO of Snowflake, made Wednesday morning during his keynote address at The Knowledge Graph Conference in New York.
The topic of his talk, “From the Modern Data Stack to Knowledge Graphs”, brought together approximately fifty years of database technology in a new form.
The main story is this: five companies have created modern data analysis platforms, Snowflake, Amazon, Databricks, Google and Azure, but these data analysis platforms cannot perform business analysis, including, most importantly, present the rules. which are the basis of compliance and management.
“The industry knows this is a problem,” Mulia said. The five platforms, he said, representing the modern data stack, have allowed a “new generation of these very, very important data applications” to be built. What we can’t do effectively, I’d say the number one problem customers have with all five of these platforms is management. ”
Also: Ethernet creator Metcalfe: Web3 will have all kinds of “network effects”
Muglia, who runs Microsoft’s SQL Server business, among other achievements in his 30-year career in database building, provided illustrations of business rules that data platforms cannot model.
“So if you want to run an application to say ‘Hey, tell me all the resources Fred Jones has access to in this organization,’ it’s a difficult application to write,” he said. “In fact, this is a query that probably can’t be executed effectively in any modern SQL database if the organization is very large and complex.”
The problem, Muglia said, is that algorithms based on structured query language, or SQL, cannot make such complex “recursive” queries.
“Many generations of algorithms have been built, all around the idea of binary,” Mulia said. “They have two tables with the key to link the two together, and then you get a set of results, and the Query Optimizer accepts and optimizes the order of these mergers – binary join, binary join, binary join!”
Recursive problems such as Fred Jones’ solutions, he said, “cannot be effectively solved with these algorithms, period.”
The right structure for business relationships, as opposed to data links, Mulia said, is a graph of knowledge.
“What is the graph of knowledge?” Mulia asked rhetorically. He proposed his own definition of what can sometimes be a mysterious concept. “Knowledge graphics is a database that models business concepts, the relationships between them, and related business rules and constraints.”
Muglia, now a member of the launch board Relational AItold the audience that the future of business applications will be knowledge charts built on data analysis, but with the reverse that they will use relational calculus, going back to the pioneering relational database EF Codd.
“Go back to the beginning,” Mulia said. “What is the basic algorithmic ability of relational technology, what can be done with it?”
“If we look at what Code created in the 1970s, the theorem says that relational algebra and relational queries are exactly equivalent in expressive power – so interesting! I’ve always known it was interesting, and now I know why.”
Using Relational AI technology, in particular the technology developed by founder and CEO Molham Aref to optimize queries, Mulia said, combined with knowledge graphs, it is possible to bring the same relational algebra into the organization of business concepts.
“If we move to relational knowledge graphs, we now go to the basics of relational calculus and publish relational computational expressions that are disordered and contain business rules and constraints.”
Work to “fundamentally redefine the way we can use relational algebra,” Mulia said, has been going on for about 20 years, but intensified in 2010 with Aref’s work at many universities and companies, with hundreds of publications published in the topic, he said.
“It was an incredible collaboration between research communities around the world,” he said. “Materialize uses it in a fundamental way within the materialization they perform. LinkedIn embraces some of that through a graphical database they use called Liquid.”
The relational knowledge graph introduces a new language called Rel, although “SQL remains important,” Mulia said, “SQL will not disappear,” as it serves as a kind of ramp to the new world of relational knowledge graphs.
“I can even say that the best days of SQL are ahead of him.”
Muglia foresees “amazing, incredible potential in terms of software development and all sorts of things come from that,” he said of the relational knowledge chart. “When technology matures – and I want to focus on the fact that it is not yet mature – but as technology matures, we will see things we can do with it that we could not even dream of.
This includes much more use of machine learning in business analysis, as the name Relational AI suggests. Besides, the business model, Mulia said, will no longer be “just something we put on a whiteboard that the engineer has to look at to write some Java code against it.”
“The model,” Mulia said, “is becoming the program so business analysts can get involved and make changes to data structures.”
“Think of the thousands of people involved who know about business – think about it!”
Relational AI technology is currently “a pretty white glove,” he said. “We have a number of organizations that we do limited testing with,” and the company hopes to “open up for self-service” next year with a broad beta for developers so people can sign up and start using the system. “It’s the same trajectory,” he said, as he did when he ran Snowflake in 2014 and released early versions of his code.
“We are at the beginning of a whole new era,” Mulia said. “It’s like the current data stack in 2013, 2014 – that’s where we are in this life cycle.
“And just as the modern data set has revolutionized analytics, I am fully convinced that knowledge graphs, and relational knowledge graphs in particular, will transform the way business works.”
The Knowledge Graph conference is in its fourth year, starting its life as a small affair in a ballroom at Columbia University in 2019. This year, after only two years of virtual procedures, the conference has become an extensive hybrid event with dozens of panels and live sessions at the Cornell Tech campus on Roosevelt Island, New York. The program lasts until May 6 .