vectorfusionart - stock.adobe.co
TigerGraph wants to bring graph database technology to the masses.
To that end, the vendor, founded in 2012 and based in Redwood City, Calif., recently unveiled TigerGraph on Microsoft Azure and took steps to make its proprietary coding language, designed to resemble SQL, part of the standard coding language for the development and querying of graph databases so it can be quickly learned and used by anyone with experience in coding.
Graph databases differ from more traditional relational databases by attempting to simplify the connection of data points while simultaneously allowing data points to connect with more than just one other data point at a time. In theory, they speed up the query process and enable users to easily extract data from disparate sources.
"Democratization of graph is our mission statement for the foreseeable future," said Gaurav Deshpande, TigerGraph's vice president of marketing. "We don't want to see customers using it like it's a fancy thing and use it just for one little project. We want to them to use it as an everyday part of their lives. Our objective is to make graph accessible and available."
Expansion in the cloud
TigerGraph emerged from stealth in 2016, and in September 2019 launched TigerGraph Cloud, the database-as-a-service version of its graph database which has both a free tier and pay-as-you-go tier. Up until just a few weeks ago, however, TigerGraph Cloud only offered support for Amazon Web Services. TigerGraph Cloud users that deployed on other clouds couldn't use TigerGraph natively, and that proved to be a barrier as TigerGraph tried to attract more paying customers.
Meanwhile, according to Deshpande, users asked for native support for their cloud and said they'd upgrade if TigerGraph could provide it. So on July 16, TigerGraph added support for Microsoft Azure, and support for the Google Cloud Platform (GCP) will be added within the next few months.
Gaurav DeshpandeVice president of marketing, TigerGraph
"We saw a lot of people coming on board, and they were all Azure people," he said. "They told us, 'We like TigerGraph Cloud, but even to start in the public cloud I cannot start with AWS -- I need to have my workload inside an Azure Cloud,'" he said. "That was the impetus for us adding Azure as an option. Azure support was specifically in response to a market force."
He added that TigerGraph conducted a survey of 300 customers who said that if TigerGraph offered native support of GCP they would use it.
Support for multiple clouds, meanwhile, is becoming commonplace among analytics vendors, and according to Mike Leone, senior analyst at Enterprise Strategy Group, is important for a relatively new vendor like TigerGraph as it attempts to expand its customer base.
"It's about bringing graph to where the data is," he said. "And for 80-plus percent of organizations, it's in multiple locations, on premises and in multiple clouds."
Beyond adding more cloud options to attract more paying customers, TigerGraph is trying to expand the use of graph databases by making its graph query language as simple to use as possible.
The vendor's query language is called GSQL, and from the start was designed to be as close to SQL as possible so that data scientists and application designers familiar with SQL would be able to easily adapt to TigerGraph and turn their data into actionable models data sets.
GSQL essentially starts with SQL, but then adds what TigerGraph calls accumulators that are designed to enable users to perform complex computations on connected data sets within graph databases in ways they're not able to within relational databases.
In addition to writing GSQL itself, however, TigerGraph is attempting to help standardize the coding language behind graph databases, and recently submitted a paper to SIGMOD -- a special interest group on data management that's part of the Association for Computing Machinery -- and presented it at SIGMOD 2020 in June.
The paper made the machinery behind GSQL public so that aspects of GSQL -- in particular the accumulators -- can be integrated into GQL, the language for graph querying being developed by the International Organization for Standardization that already oversees the development of SQL.
"It will take a few years because it's a fairly involved functionality, but we're excited that what we have learned from customers we are donating back to the graph query language and the movement for having a standardized graph query language," Deshpande said.
Leone, meanwhile, said that easing the transition to graph databases is critical for TigerGraph and its continued growth.
"Simplicity, simplicity, simplicity," he said. "Simplifying adoption of graph databases is the greatest barrier to broader adoption across the industry.
"Organizations have the data, they have the need, and TigerGraph is focused on simplifying ramp up of using graph technology. And it's not just about experts. Enabling generalists to use the technology is critical to success."
Regarding the move to make GSQL as easy for programmers to learn as possible and help create a standardized graph query language, Leone added that too will only help TigerGraph in the long run.
"Enabling a seamless and easy transition from SQL to QSQL matters for many potential customers," he said.
Just as TigerGraph's addition of support for Microsoft Azure and the publication of its coding language harken back to the vendor's goal of making graph technology accessible to the masses, its roadmap is similarly shaped by that aim.
Future development, Deshpande said, will center around adding more in-database machine learning capabilities and more no-code capabilities so that business users can do increasingly complex queries without needing to learn GSQL.
TigerGraph sees itself becoming the "de facto standard" for relationship analysis and as a vendor that simplifies complex relationships for users, Deshpande said.
"Expect us to do more on that front, both driven by customer demand, as well as our own mission of making graph available to everybody in the world," he said.