Oracle Autonomous Database – is it truly self-driving?
Oracle generated a lot of buzz prior to Oracle OpenWorld 2017 last September with their announcement of the world’s first self-driving database - Oracle Autonomous Database. However, not many details were released at announcement time. Now that the first Oracle Autonomous Database service, Autonomous Data Warehouse Cloud, was finally released in March 2018 let’s take a look at what Oracle provides.
What is the Autonomous Database?
Oracle Autonomous Database is comprised of Oracle Database 18c Enterprise Edition (with all database options and management packs included from a software packaging perspective), Oracle Exadata X7-2 as the deployment platform, AND the Oracle Cloud Automated Database and Database Operations components (scripts, best practices, procedures, etc.) (1).
By our analysis, Oracle Database clients CANNOT replicate the Autonomous Database services in their own Oracle environments by virtue of its architecture: namely that Oracle Cloud Automated Database and Database Operations is a primary component. For now, it appears that would relegate availability to the Oracle Public Cloud. In contrast, IBM Db2 has had built-in autonomic capabilities for over ten years that can be used by clients in their own non-IBM-controlled environments.
Is it a revolutionary breakthrough in database technology in the public cloud?
At its heart, we see Oracle Autonomous Database as a set of fully managed database services that are optimized for a specific workload in the public cloud. IBM Cloud has offered fully managed database services with its Db2 on Cloud and Db2 Warehouse on Cloud offerings for over two years. Although Oracle Autonomous Database currently (May 2018) offers a data warehouse service ONLY, IBM also provides a fully managed solution for operational (transactional) workloads today(2) .
Even the Oracle Autonomous Data Warehouse Cloud allowing database compute resources and storage resources to grow independently of each other is not new, IBM Cloud offers this type of flexibility for both Db2 on Cloud and Db2 Warehouse on Cloud.
Some facts about Autonomous Database and its 99.995% availability claim of less than two minutes of downtime per month with no exclusions either planned or unplanned outages.
Oracle offers two configuration choices for Autonomous Database: Enterprise and Mission Critical. However, we found the 99.995% availability is ONLY true for the Mission Critical configurations which could cost clients up to 2X the monthly subscription costs. This increased cost estimate is based on the requirement of a standby database which will double the infrastructure and storage and therefore could double the cost, if there are no additional discounts.
The Autonomous Database Enterprise configurations offer the same 99.95% availability that the IBM Cloud and its Db2 on Cloud (operational workloads) and Db2 Warehouse on Cloud (data warehouse analytics workloads) services provide today.
Let’s compare Oracle Autonomous Data Warehouse Cloud with Db2 Warehouse on Cloud Flex Performance
Oracle Autonomous Data Warehouse Cloud utilizes Oracle Exadata X7-2 systems as its deployment platform, and only dedicates OCPUs (according to Oracle’s definition an OCPU is equivalent to an Intel Xeon physical core) and a single PDB (Pluggable Database) to each Oracle Cloud client. (3) The memory, flash cache, and other resources on an Oracle Exadata system are not dedicated. Oracle will decide who and how many other clients share the PDB resources available within a single Exadata machine(4). Autonomous Data Warehouse Cloud does not advertise an included capability to incorporate other relational data sources or non-relational data sources with the data in the cloud data warehouse without moving that data to the Oracle Cloud (5).
With a multitenant configuration within Autonomous Data Warehouse Cloud, you will not even know how many additional clients are sharing your Oracle Exadata machine. Although Oracle can control data access and PDB separation from a security perspective, there is no advertised purchasing option which allows you to dedicate memory and flash cache to specific PDBs (only OCPUs are dedicated).Even with Database Resource Manager to control database resources at the PDB level, we believe that an individual PDB CANNOT be guaranteed consistent performance between off-peak and peak usage periods of the Oracle Exadata appliance. Clients with strict SLAs for their production data warehouses will have to ask themselves if this type of data warehouse implementation (multitenant) is acceptable. Only time and client experiences with Oracle Autonomous Data Warehouse Cloud can answer this question.
Db2 Warehouse on Cloud provides dedicated hardware compute resources, both cores and memory, to each Db2 Warehouse on Cloud Flex Performance cluster. This type of approach is engineered to be more consistent than shared-resource models such as Autonomous Data Warehouse Cloud. Whereas shared resource models may cause you to compete against other clients for those resources at peak times, slowing performance, the dedicated resources can be put towards your tasks at the same level of performance on a reliable basis. There will be total separation between different client databases (6),(7). IBM gives you a dedicated database environment versus a dedicated PDB within a group of PDBs. Db2 Warehouse on Cloud provides an INCLUDED data virtualization capability to allow both relational and non-relational (Hadoop) data sources from various vendors to be used “as-is” where the data currently resides in combination with the data within your cloud data warehouse.
IBM’s Db2 managed cloud offerings deliver. They provide dedicated compute (including memory) and a data virtualization layer that allows both relational and non-relational data sources to be used in a logical data warehouse scenario. Unlike Oracle Autonomous Database, IBM provides a true hybrid data management solution that can exploit data wherever the data resides, including other vendors databases AND provide the consistent performance that clients demand for their production data warehouses.
If you have additional questions about the differences between Oracle and IBM’s solutions, reach out to one of our experts and they will be happy to answer.
(1)“Oracle’s Autonomous Database 18c: An Initial Assessment”; 21 Nov. 2017; http://www.oracle.com/us/corporate/analystreports/ovum-automomous-db-4175962.pdf
(2)“Oracle’s Autonomous Database 18c: An Initial Assessment”; 21 Nov. 2017; http://www.oracle.com/us/corporate/analystreports/ovum-automomous-db-4175962.pdf
(3) Oracle Cloud Compute Page; https://cloud.oracle.com/en_US/compute
(4) Oracle Multitenant White Paper; June 2013; http://www.oracle.com/technetwork/database/multitenant-wp-12c-1949736.pdf
(5) Oracle Multitenant White Paper; June 2013; http://www.oracle.com/technetwork/database/multitenant-wp-12c-1949736.pdf
(6) Oracle Multitenant White Paper; June 2013; http://www.oracle.com/technetwork/database/multitenant-wp-12c-1949736.pdf
(7) “Understanding performance interference in multi-tenant cloud databases and web applications”; 06 Feb. 2017; https://ieeexplore.ieee.org/document/7840933/