Reg comments1

Oracle is red, violets are blue, we hope you'll integrate biz analytics in our cloud soon

Blue, soon, close enough

Oracle acrobatics in the cloud

At Oracle, the future is cloud. If a product is not in the cloud, it's toast. It's that simple. The command has come down from on high to Oraclers. Be in the cloud or be finding a new job.

Here's a great example of that diktat in full effect. Oracle has a thing called the Oracle Data Integrator Enterprise Edition. Well now you can get it in the cloud.

It's called the Oracle Data Integrator Cloud Service, and basically it's a way to centrally perform extract-load-transform operations on large amounts of information. The service can create a new database from information pulled from storage – whether that data is in Oracle's public cloud, a rival public cloud, or a private on-site repository – and then run code over that new database.

Of course, Oracle hopes you'll do all that processing in its cloud.

We're told ODICS is standards-compliant: for example, it can access Hive, HDFS, HBase, and Sqoop systems; and use SQL, HiveQL, or bulk-transfer scripts, to retrieve records. Speaking to The Register, Jeff Pollock, vice president of product management at Oracle, said ODICS will provide the full functionality of Oracle's Data Integrator Enterprise Edition.

The key thing here is that ODICS is an extract-load-transform (ELT) affair whereas you may be more familiar with extract-transform-load (ETL) tools that retrieve information, process it and then store it back in a new database. "ETL tools bring the data to the engine and transform it in the engine, we're bringing the transformation to the data," said Pollock.

It sounds a little six of one, half a dozen of the other. The difference is explained well here: ELT is great for pulling in information to a new store and then transforming it there, rather than transforming it at the source and then saving the results elsewhere. If you're operating on a production system, you'll want to copy records off the servers and then process them in a separate backend, rather than run the processing on the live user-supporting gear. You don't really want to run analytics or report generation on systems busy handling customers when you can work on a copy elsewhere.

Customers can run ELT processes within their own data warehouses, without that information venturing beyond their firewalls, or do the processing in the cloud, having copied out the data from other stores and collated it off-premises. According to Pollack, Oracle hopes the "next great wave" in technology will involve business reporting and number crunching in the cloud – which will mean running processing apps off premises.

The migration of data to the cloud for background processing "will drive the growth for these types of data management tools in the cloud – we will see that happening over the next months," said Pollock.

ODICS is charged on a per-connection basis, with a maximum usage of up to 100,000 5MB messages per connection per day. Non-metered service costs $650 per connection (minimum six connections per month), while metered pay-as-you-go service costs $1,300 per connection (minimum two connections per month). ®

Biting the hand that feeds IT © 1998–2017