Hey Presto! Teradata admits its vision is dead by hooking QueryGrid analytics platform up to rival data warehouses
Snowflake? We can query that
Updated Enterprise data warehouse stalwart Teradata has capped a difficult year with an update to Teradata QueryGrid, which promises to connect customers to a vast array of new data sources – a decidedly underwhelming move, according to some.
Teradata has partnered with Starburst Data to integrate a new Presto connector so that users of Teradata's Vantage analytics platform can access and query a gamut of cloud and on-premises data sources.
Document store database MongoDB, Databricks Delta Lake, columnar database Apache Cassandra, as well as cloud data warehouses Amazon Redshift and Google BigQuery were listed by Teradata as the data sources at users' fingertips as a result. In its publicity material Teradata somehow forgot to mention that the tool also connects them to Snowflake, the headline-grabbing cloud data warehouse stock-market smash of the year.
The omission could be a sign of sensitivities at San Diego-based Teradata, one of the companies that defined the data-warehousing market from the 1980s onwards.
The idea was to create a single store of business data that could be efficiently queried by thousands of users, with the aim of helping businesses make better-informed decisions. It always suffered from the tendency of users and departments to squirrel way data in spreadsheets on desktops and local servers. With the advent of cloud, the problem has mushroomed.
Another cloud in the sky...
By giving access to many other data sources, including rival data warehouses, Teradata is as much as admitting that its vision is dead. It has positioned itself as a "cloud data analytics platform company" for a while in any case.
Open-source Presto is a SQL query engine written in Java. Its distributed architecture was originally developed at Facebook back in 2013 to query data residing in Hadoop.
Its partner in the new effort, Starburst Data, was rolled out of Teradata in 2017 to help support and develop Presto, and has former Teradata staffer Justin Borgman as co-founder and CEO.
Teradata said the update to QueryGrid gives companies access data in hybrid and multi-cloud environments and pushes processing down to where the data resides, reducing data movement and therefore decreasing costs.
The company is crawling towards the end of a difficult year, which has seen a change in CEO, disappointing financials, and industry attention focused on cloud-native data warehouse providers, especially Snowflake, which bust $120bn in market cap last week.
Clara Angotti, president of enterprise data and cloud migration services at cloud start-up Next Pathway, said the move to extend QueryGrid was an example of Teradata trying to find new sources of revenue.
"The concept of dynamic access to data across multiple platforms is useful to customers in large and complex environments. However, what Teradata is trying to do is cement themselves as the centre of these data ecosystems through their new service.
"Optimising egress costing is interesting, but again, customers will be increasingly dependent on Teradata in these environments. I'm sceptical if enterprise customers will see value in this," she said.
An alternative to Teradata's approach would be a method that sought to optimise data models in the cloud when migrating from legacy applications and data warehouses to help make consumption easier and more efficient, Angotti added.
Life might be tough for Teradata right now, but there is still a chance it could get through this particular blizzard. Philip Howard, research director at Bloor Research, said the QueryGrid update was "clearly a data virtualization play even if they are calling it a fabric."
QueryGrid was originally built for seamless queries across Teradata data warehouse and data exploration platform Aster, he added. "Now they're extending it to other environments, and leveraging Presto. Since everybody offers data virtualization nowadays, it is partly a question of catch-up and partly a question of how good and how broad data virtualization actually is."
Here the firm may achieve better results than most competitors, he said.
"Teradata is having a hard time. However, we have heard multiple stories about companies unhappy with Snowflake and some even moving off it. Teradata has been here before, with the launch of [hardware-optimised] Netezza [founded in 1999] and weathered the storm. I wouldn't bet against them doing the same again."
Updated on 18 December at 16.04 GMT to add:
Following publication of this article, Terradata sent us a statement:
"Teradata believes in a single version of the truth. That vision isn’t dead; but sure, our approach has changed."
The company said it understands that customers' data is held on-prem and in multiple clouds and they may want to run ananlytics on "all of it".
"As competitors’ analytical silos multiply across the organization, they continue to provide value at the micro level, but drive up cost, IT complexity, adversely impact time to value and ultimately, don’t give a complete picture of the business.
"So no, we don’t shy away from providing connections to our competitors’ products, hybrid cloud or multi-cloud environments, because it’s essential to helping our customers realize the benefits of integrating of their data: cost savings, efficiency, optimization and ultimately, business opportunity."
Got that? Good. ®