The place Teradata might go along with its information lakehouse

Final week Teradata provided  its long-awaited response to the data lakehouse. As George Lawton recounted final week on VentureBeat,, Teradata has all the time differentiated by stretching the capabilities of analytics, first with massively parallel processing by itself specialised machines, and extra not too long ago, with software-defined home equipment tuned for variations in workloads from compute-intensive to(enter/output operations per second (IOPS)-intensive and/or high-concurrency. And for the reason that acquisition of Aster information over a decade in the past, Teradata morphed from fixing huge analytics issues to fixing any analytics downside with a various portfolio of analytic libraries stretching SQL to new areas resembling path or graph analytics.

With the cloud, we’ve been ready for when Teradata would totally exploit cloud object storage, which is the de facto data lake. So the twin bulletins final week of VantageCloud Lake Version and ClearScape Analytics had been logical subsequent steps on Teradata’s journey to the information lakehouse. Teradata is lastly making cloud storage a first-class citizen and opening it as much as its extensive analytics portfolio.

However not like Teradata’s earlier strikes to parallelized and polyglot analytics the place it led the sector, this time with the lakehouse, it has firm. The announcement won’t have talked about the lakehouse phrase, however that’s what it was all about. As we noted several months back, nearly everybody within the information world from Oracle, Teradata, Cloudera, Talend, Google, HPE, Fivetran, AWS, Dremio, and even Snowflake has felt compelled to answer Databricks, which launched the information lakehouse.

Teradata’s path to the information lakehouse

Nonetheless, Teradata approaches the information lakehouse with some distinctive twists and they’re all about optimization. Teradata’s secret sauce has all the time been about extremely optimized compute, interconnects, storage and question engines, together with workload administration designed to run compute assets as much as 95% utilization. When commodity {hardware} received adequate, Teradata launched IntelliFlex the place efficiency and optimizations might be configured by way of software program. The aptitude to optimize for {hardware} not-invented-here opened the door to Teradata optimizing for AWS, and down the street, the opposite hyperscalers.

Teradata launched VantageCloud a yr in the past, and late final yr ran a 1,000+ node benchmark that no different cloud analytics supplier has to date matched. However this was for a extra standard data warehouse utilizing customary block storage.

The cog within the wheel for making the lakehouse occur was creating a desk format for information sitting in cloud object storage. That permits all of the niceties related to information warehouses, resembling ACID transactions, which is essential to making sure consistency of information, extra granular safety and entry controls, and uncooked efficiency. Databricks fired the primary shot with Delta Lake, and extra not too long ago, different suppliers from Snowflake to Cloudera and others have embraced Apache Iceberg, with the widespread thread being that that is all based mostly on open supply expertise. For Lake Version, Teradata went its personal approach with its personal information lake desk format, which the corporate claims delivers superior efficiency in comparison with Delta and Iceberg.

The opposite aspect of the lakehouse coin is software program. Apart from its SQL engine, which has been designed to deal with giant, advanced queries that may be part of as much as tons of of tables, Teradata has a big portfolio of analytic libraries that run in-database. This has been considered one of Teradata’s best-kept secrets and techniques. Largely the legacy of the Aster Knowledge acquisition over a decade in the past, these analytics had been specifically tuned to use the underlying parallelism, they usually went properly past SQL, encompassing features resembling n-Path, graph, time collection evaluation, and machine studying, all of which accessed by way of SQL extensions.

Formally branding the portfolio as ClearScape Analytics, Teradata is lastly drawing consideration to the truth that it’s a holistic analytics platform and never merely an information warehouse, information lake or lakehouse. As a part of the announcement, Teradata beefed up the time collection and MLOps content material. However after we take care of the information lake, information scientists are very opinionated on selecting their very own languages or instruments. And so, VantageCloud may even help a ring-our-own-analytics choice for these preferring to write down Python and work from Jupyter notebooks or their very own workbenches, and at the moment has integrations with Dataiku, KNIME and Alteryx. ClearScape analytics will likely be out there, each for VantageCloud Lake Version and the usual Enterprise Version.

Lake Version and ClearScape Analytics are promising begins for Teradata as information lakehouse. There’s little query that Teradata’s scale and help of polyglot analytics made lakehouse a query of when, not if. And branding the analytics portfolio is greater than only a advertising train, because it lastly shines the highlight of what had been a best-kept secret: Teradata’s differentiation goes past the optimized SQL engine and infrastructure to incorporate analytics optimized for that engine. VantageCloud takes the analytics portfolio full circle by unleashing the portfolio on cloud object storage, and with usage-based pricing, doubtlessly opens up the portfolio for extra discretionary workloads in comparison with the times when prospects had been working on-premises with agency ceilings on capability.

A want checklist for Teradata

That leaves our want checklist for what Teradata ought to do subsequent. In abstract, we wish to see Teradata enterprise additional out of its consolation zone to attract new audiences of customers. Admittedly, with the lakehouse, the problem will not be distinctive to Teradata, as Databricks, for instance, seems to attract in enterprise analysts whereas Snowflake courts information scientists.

To attract that new viewers, Teradata ought to decrease entry boundaries and put open supply on a extra degree footing with its proprietary setting. With Lake Version, Teradata has dramatically lowered its entry pricing to $5,000/month. That could be a marked drop from the six- and seven-figure annual contracts that Teradata prospects sometimes pay, however we’d wish to see Teradata go additional with a freemium providing that enables new customers to kick the tires. Heck, even incumbents not recognized for low cost pricing like Oracle have embraced free tiers.

As for open supply, there are a pair pathways that we’d wish to see Teradata additional develop. The primary is drawing non-Teradata customers to ClearScape Analytics by way of optimized APIs to open supply Delta and/or Iceberg information lakes. Whereas efficiency won’t be on par with Teradata’s personal information lake desk format, it might be made “adequate.” 

Conversely, we’d wish to see parallel efforts with so-called BYO analytics, drawing the Python crowd by way of optimized APIs with Teradata’s personal information lake desk format. As an illustration, we want to see Teradata workforce up with Anaconda for juice efficiency of the Conda Python library portfolio, a lot as Anaconda is already doing with Snowflake. On the finish of the day, it’s all concerning the analytics.