While Teradata Corp.’s recent acquisitions should leave little doubt that the data warehousing giant is serious about distributed Hadoop Big Data, the company is also demonstrating that it hasn’t forgotten about the big integrated appliances that got it this far. The company today used its European user conference to introduce a new integrated warehousing machine that that doubles data capacity and halves the physical footprint with a design that’s optimized for fast in-memory processing.

The company also announced enhancements to its QueryGrid cross-platform query engine and added a software-defined warehousing capability that supports multiple isolated instances on a single processor with consolidated administration and reporting.

Teradata is positioning its new Data Warehouse Appliance 2800 as a “data warehouse, data mart, disaster recovery system, or as an analytical sandbox for testing and development.” The machine combines dual Intel Xeon 14-core processors with vector processing capabilities in Intel’s new Haswell technology, updated DDR4 memory components, a Linux operating system and integrated storage to achieve an average 15% performance improvement, according to Imad Birouty, director of product marketing. The reason the performance boost isn’t greater is that Teradata switched from hardware to more-flexible software compression in this engine in this unit, which impacts CPU performance. “If you’re not doing any compression and you’re also doing operations that benefit from in-memory optimization, you system is going to go a lot faster,” Birouty said.

Teradata also tapped technology from storage subsystem maker Dot Hill Systems Corp. to fill previously unused rack space with SAN drives and squeeze more storage into the same space, improving performance and reducing sitze. Each machine can accommodate up to 12 nodes per cabinet with up to 512G-bytes of memory per node with four storage configurations available ranging from 45.8T-bytes to 320T-bytes. The system is scalable to more than 54P-bytes using 1200G-byte drives

QueryGrid is intended to unify the multitude of data warehousing platforms that many enterprises now support. The technology enables a user to submit a single query for execution across multiple destinations “without your knowing or caring where it executes,” Birouty said. Teradata does this with use of connectors, and with today’s announcements is updating the roster to support Cloudera, Inc. CDH 4.3 and 5.1, Hortonworks, Inc. 2.1 and Teradata’s namesake database as well as its Aster Discovery Platform. The company also added a new connector for MapR Technologies, Inc.’s Hadoop distribution.

Finally, the company is making it easier for users to segment Teradata instances into walled gardens in which each instance acts as a standalone data warehouse with its own security parameters and service level agreements. In effect, the company is virtualizing its data warehouse into silos. “Users can’t see anyone else on other systems, and their performance is guaranteed,” Birouty said.

Use cases include multi-tenancy applications with multiple business units or departments, compliance scenarios in which data between different countries must be kept separate, service provider deployments and data mart consolidation projects.

While individual warehouses are logically separate, administrators can run data rollups and consolidation reports across multiple systems. “We specifically chose to market his announcement in Europe because the regulatory environment makes this a desirable feature,” Birouty said.