Follow Datanami:
November 4, 2022

Teradata Taps Cloudian for On-Prem Lakehouse

Teradata has reinvented itself as a cloud-first data warehouse provider with its Vantage offering. But some customers don’t want to run analytic workloads in the cloud, which is why the data warehousing giant this week announced a partnership with object storage provider Cloudian that will allow Vantage customers to keep their data on premise.

According to Cloudian, the combination of its HyperStore storage layer with the IntelliFlex and VMware editions of Teradata Vantage will deliver a cloud-like data lakehouse in support of analytic workloads running in on-prem and hybrid cloud environments.

Running analytics in the cloud works for many customers. But for others, issues around cost, security, sovereignty are proving substantial obstacles to running in the cloud, says Larry Meese, Cloudian’s vice president of products and solutions.

“Not all data can go to the cloud due to regulatory, compliance and/or legal concerns/constraints,” Meese tells Datanami via email. “For massive data sets at large scale, public cloud can quickly become cost-prohibitive for many organizations.”

HyperStore is an S3-compatible object storage system that can scale to petabytes running on a cluster of X86 servers. Just like public cloud-based deployments of Vantage, on-prem deployments would use separate clusters, one to house the data in HyperStore, and another to run the Vantage query engine.

Interest in on-prem deployments is high among new and existing customers, says Meese. Clients in financial services, healthcare, education, service providers, telcos, and all levels of government have shown an interest in on-prem deployments of data warehouses using an S3 storage tier.

A hybrid cloud setup is another deployment option for Cloudian-Teradata customers, he says. There are several hybrid setups that customer might use, including running Vantage in the cloud while keeping the data storage on-prem; running Vantage in the cloud while storing data on-prem and in the cloud simultaneously; and running both the processing and storage tiers on-prem, but using the public cloud to store archived data.

Giving customers options helps to keep their costs down and meet performance SLAs. “Obviously, latency will be increased for queries that have to traverse an external network to reach the required data, so that will mean queries will take longer to run in those configurations,” Meese says. “This is an important design decision and tradeoff that needs to be considered for customers exploring a hybrid configuration.”

The joint Cloudian-Teradata on-prem setup is not yet in production, as Cloudian just recently was certified, Meese says. “However, we have a handful of large, existing joint-enterprise customers that will be deploying in the near future,” he says. “We will have a reference architecture available later in November.”

Cloudian has devised this sort of setup before. In June, it inked a deal with MicroFocus (the owner of Vertica) to store data on-prem in its S3-compatible object store, and in July, it announced a partnership with Microsoft to provide an on-prem storage option for data to be processed using the Microsoft Azure stack.

Related Items:

Object and File Storage Have Merged, But Product Differences Remain, Gartner Says

Teradata Unveils New Data Lake, Advanced Analytics Offerings

Teradata Rides Cloud Wave to Two-Year High