Oracle provides database 12c with "unchristian" in-memory turbo


With the help of a new technology Oracle wants to accelerate the new version of 12c with inquiries by the factor 100. Oracle uses in-memory technologies, which are also used in the SAP competitor HANA.

Oracle will offer this in-memory technology as an option. Larry Ellison unveiled Oracle's new offer this weekend at Oracle OpenWorld. According to Oracle, this new feature should accelerate queries significantly, but adjustments to existing applications are not necessary.




"It will make Queries 100 times faster. You can load the same data into identical machines and it's 100 times faster, you get the results in the blink of an eye, "explains Ellison at the announcement of the new offer. About costs and other details about the product Oracle has so far not shared.

However, it is not enough to accelerate queries alone. "We did not want transactions to slow down by adding or changing data. We have therefore found a way to accelerate query processing and thereby at least double transaction processing rates. "

As with SAP HANA, it is not enough to simply load the data into the main memory. The architecture of the databases must therefore also be adapted. In traditional database architectures, the data is stored in rows. Fast queries can be achieved by having many columns in a few rows. However, storing data in columns increases the query speed.

Now, Oracle will offer both options, so to speak, Ellison explains. Thus, roughly formulated transactions can be stored relatively quickly in rows and analyzes or analytical software can be processed through the columns much more effective.

"We can now analyze data with unchristian speed." At least in the presentation of Ellison, who underlined this claim with a demo at OpenWorld.

Anyone who creates a table in Oracle today must also decide which columns to index. "We have replaced these analytical indices with the in-memory option. Let's forget those analytic indices and replace them with Column Store, "Ellison continues.

Companies would not have to provide the entire database with the in-memory option. It could also be accelerated individual parts of it with in-memory. And, of course, the price of the solution is cheaper. You can save hot data in the DRAM, some in Flash and others on disks.

The data would then automatically migrate to the various storage media based on
usage patterns, assures Ellison. And then you only have to pay for the capacity of the disk. Oracle already has very elaborate technologies here.

Companies could also start such projects at any time. No features need to be adjusted, data reloaded or migrated, Ellison assures. The improvement results solely from the adjustments in the structure of the database.

The new multienancy options for cloud applications also run with in-memory. According to Oracle, Oracle does not drive two versions of the same datasets. It only stores the transactions for the row-based storage, but not a second time for the columns.

To make the most of this technology, Ellison also introduced the new M6-32 hardware , codenamed Big Memory Machine.

LihatTutupKomentar