Teradata is not only its size and performance, but also advanced, designed to handle very complex functions such as recursive queries, implicit joins, unique syntax and custom logic for parallelization of workloads. It has long been highlighted by the SQL engine. As a result, Teradata has long positioned itself for organizations with the most difficult analytical challenges. And finally, Vantage has aggressively embraced the cloud.

The world’s Redshifts, Synapses, Snowflakes, and Big Queries are positioning themselves as the latest paid hyperscale cloud alternatives that provide a more cost-effective alternative to traditional Teradata platforms. Functional gaps and source code and schema modification requirements are often the focus of migration.

Of course, there are startups who believe there is an answer to this.

According to Datométrie, the answer is database virtualization, not data virtualization. The approach is to insert a runtime that acts as a buffer between the Teradata SQL statement and the target cloud data warehouse. The idea is to allow Teradata customers to run Teradata queries against a variety of targets without modifying or completely rewriting existing SQL programs. Its product, HyperQ, adds Oracle to its list of database sources.

At the heart of the Datometry approach is a single hypervisor that emulates SQL database calls on the fly. Internally, it breaks down these complex calls, stored procedures, and macros into atomic operations that the target data warehouse needs to understand. For example, a recursive query used to query a nested or hierarchical data structure is transformed on the fly into a series of simple individual calls to the target, with intermediate results stored in a temporary table maintained by the hypervisor. Will be done. These operations can be complex and provide a rule-based queue that matches existing rules executed at the source. Provides JDBC and ODBC APIs for BI and ETL tools.

Of course, Datometry isn’t the first person to say “don’t change the program”. There are SQL translators, but Datométrie claims that they are often insufficiently efficient. They estimate that the code converter should handle around 60-70% of all workloads. The traditional workaround was to add non-SQL code to the application to compensate for the difference between Teradata SQL and the target SQL database. Likewise, custom data types and structures are often overlooked by cloud database schema migration tools.

Can Datometry handle all the peculiarities of Teradata SQL? The company claims to cover 99% of its Teradata workloads. Of course, it costs money. Datometry virtualization layer adds 1-2% of the overhead, but as EPA rates mileage depends on workload. The company says it’s a small price to pay compared to the overhead of maintaining the code from the SQL code and schema conversion tools.

Datometry performed its first proof of concept using SQL Server on an on-premises HPE Superdome machine about four years ago, then began supporting Azure Synapse and Google BigQuery in the cloud. As mentioned above, we just announced a preview of Oracle. It’s important to note that Datometry hasn’t targeted Amazon Redshift or Snowflake yet, so it’s still suspended.