TigerGraph Upgrades GSQL
July 24, 2020
brings modern aggregation support to graph analytics, and is a property
graph query language designed for SQL users. The company’s current
advances with GSQL make this innovative query language even more similar
to SQL, the standard language for storing, manipulating, and retrieving
data within a relational database. The result: A lower barrier to entry
and easier adoption for SQL users as they chart their advanced graph
analytics course. TigerGraph’s work with GSQL has garnered industry
recognition from SIGMOD, the premier international database conference.
“GSQL’s innovation, speed and scalability boost its ease of use within
the SQL community, delivering a SQL-like experience with graph
analytics,” said Dr. Yu Xu, CEO, and founder, TigerGraph.
“Simultaneously, we’re working to help enterprises – from banks to
healthcare companies – uncover meaningful, actionable, real-time
insights from connected data that can improve people’s lives. GSQL’s
acknowledgment by SIGMOD, the renowned data management academic
conference, validates its influence on the graph database industry as
well as the database community at large. Our work in advanced graph
analytics is ongoing and ever-evolving, and we remain dedicated to
supporting inventive customer applications while innovating our product
Database experts have recognized GSQL’s power for efficient expression
and execution of graph algorithms and graph analytics. The GSQL-related
advancements complement TigerGraph 3.0, which delivers several “no code”
advanced graph analytics features. These “graph for all” features allow
data scientists and business users to move data from the relational
database to graph database and build advanced analytics patterns by
drawing -- without the need to write complex queries. TigerGraph 3.0
extends GSQL even further with rich pattern matching with analytics and
additional support for distributed GSQL.
TigerGraph’s GSQL builds on SQL functionality and extends it with a
unique construct called an “accumulator.” Accumulators allow users to
perform complex computations faster on connected datasets. The Visual
Query Builder feature in TigerGraph 3.0 allows users to employ
accumulators to aggregate without writing code. TigerGraph's
accumulators are more powerful and richer than traditional aggregation
functions in SQL; as dynamic data objects, with parallel processing like
MapReduce and Spark, they are an integral part of modern graph
TigerGraph’s Chief Scientist and distinguished researcher, Dr. Alin
Deutsch, presented the GSQL language paper at SIGMOD 2020, outlining the
massive benefits for analyzing connected datasets with graph analytics.
TigerGraph’s GSQL will support GQL, the upcoming ISO language for
property graph querying currently being created by the same
international working group that looks after the SQL standard.
TigerGraph’s language standard team -- which also includes VP of
Engineering Dr. Mingxi Wu and Head of Product Strategy and Developer
Relations Dr. Victor Lee -- is actively involved in the project. The
team will work to ensure that GQL and TigerGraph’s unique analytical
capabilities are compatible.
The research paper presents the following GSQL-related design
features, which are unique in the industry:
includes three features which synergize to form a modern graph
analytical capability and address an important class of iterative graph
algorithms – algorithms that are left out by other current and proposed
languages. These three features are default shortest-path semantics of
pattern matching, accumulators as vertex-attached run-time attributes
and global query states, and control flow primitives.
GSQL is unique in the industry due to its adoption of shortest-path
semantics as the default pattern matching semantics. Shortest-path
semantics is currently being discussed in the standard body.
GSQL’s aggregation support -- based on the concept of accumulators --
subsumes SQL-style aggregation and allows single-pass traversals that
compute in parallel multiple aggregations, even on disjoint criteria.
The paper quantifies experimentally the significant speedup of
accumulator-based vs. SQL-style aggregation for graph traversal queries.