A Veteran DBA’s Insights On Oracle’s New Autonomous Data Warehouse
Jim Czuprynski, an experienced database architect and published author on database troubleshooting, has spent a month hammering on the Oracle Autonomous Data Warehouse. His insights are worth noting, not only because Oracle Executive Chairman and CTO Larry Ellison has called the new “self-driving” technology one of the “most important things we’ve ever done,” but also because the leap to autonomous computing has many DBAs worried about losing their jobs.
Czuprynski’s three-word summary of the new cloud software: “No more knobs.”
Czuprynski’s longer assessment is that Oracle Autonomous Data Warehouse delivers on its broad promise of automating much of the manual work that has long gone into creating and running a data warehouse. Specifically, it virtually eliminates human errors; it’s continually updated and patched; it can be scaled up and down so that customers pay only for what they use; and it frees up IT pros to create new functions rather than just maintain and monitor databases.
Oracle now offers both an autonomous transaction processing database and an autonomous data warehouse as cloud services.
“Fewer knobs to turn isn’t a bad thing. The fact that there’s less for you to screw up is good,” says Czuprynski, whose books include Oracle Database 12c Release 2 Testing Tools and Techniques for Performance and Scalability and PDB Me to Oracle Cloud Pocket Solutions Guide: A Lazy DBA’s Guide to Mastering Multitenant Features on Oracle Cloud. Even more important than eliminating human error, he says, DBAs “can now focus on the things that they never had time to work on. They can help developers build better apps. We can finally get the DBA out of the back room and get you to the forefront of design.”
DBAs are worried about what this transition means for their jobs. Czuprynski, whose day job is Oracle enterprise architect at ViON Corporation, understands that fear. But he maintains that DBAs who have been embracing new database technologies, such as visual interfaces and advanced monitoring tools, will see the opportunity in taking the next step of automating the database maintenance work.
“A lot of DBAs are scared of autonomous,” Czuprynski says. “But I see it as a force multiplier.” Czuprynski will present two sessions at Oracle OpenWorld, October 22 to 25 in San Francisco, including: “A Start-to-Finish Case Study of Oracle Autonomous Data Warehouse Cloud.” Czuprynski and his co-presenters will go into detail about their experience with Oracle Autonomous Data Warehouse, but here are a few likes and dislikes from Czuprynski after his initial test drive.
A DBA shouldn’t have to spend much time monitoring Oracle’s autonomous data warehouse, but there is plenty of information to reassure IT teams that the data warehouse is performing well and has sufficient resources, he says. It has a nice interface showing how much CPU and I/O are being used, with interfaces to drill down into more performance data. “It’s a bit more minimal than what’s available with Oracle Enterprise Manager or other third-party performance monitoring tools, but it’s more than enough,” Czuprynski says, since the point is that you no longer need an experienced pro monitoring these controls.
When you do want to drill down into an SQL query, you can get a report on that query with one click as the data warehouse is running (see screen shot at right). Since Oracle Autonomous Data Warehouse is cloud-based, the infrastructure can automatically adjust to meet changing workload demands, letting DBAs focus more on the front-end work of data architecture and building better systems and queries, Czuprynski says.
With the infrastructure automated, “if an SQL statement is still not performing, maybe it’s because the SQL statement sucks,” he says. Improving SQL statements is a better way for database pros to spend their brain cycles than managing storage, or worrying about making changes that might make a single query run faster now, but have a deleterious effect on a dozen other queries later. “And one big problem with turning knobs to hopefully improve performance is that a lot of times, the DBA forgets to turn the knob back to its original setting once the crisis has been addressed,” Czuprynski says.
Czuprynski notes that DBAs will have to sort out the best approach for loading their data. One way is using SQL Loader, which can be invoked via command-line scripts or through Oracle’s free GUI SQL Developer tool. “It does take additional storage space to retain the SQL Loader input files, but that can be easily overcome by using Oracle Cloud-based storage, which is relatively cheap for retaining files while they’re being loaded,” he says. Another option is Data Pump, though Czuprynski says the syntax for performing data imports requires pretty close attention to detail when specifying the exported source file names. Oracle also has recently added the option to leverage GoldenGate for synchronizing data between a source database and the destination Autonomous database, but Czuprynski hasn’t evaluated that option yet; however, he expects that GoldenGate will be more useful for the Oracle Autonomous Transaction Processing Cloud than for Autonomous Data Warehouse.
Overall, Czuprynski urges DBAs to embrace automation and try Oracle’s autonomous database technology. Czuprynski, who has taught more than 2,000 DBAs in his prior career as an Oracle University instructor, emphasizes that those classes focused a lot on how to best tune databases and data warehouses for top performance—some of the very work that Oracle’s autonomous tools will replace.
“When you really look at what it’s doing, it’s everything I tried to teach my students in the Oracle Database Performance and Tuning course that they were supposed to do, but maybe didn’t have the time or bandwidth to do,” he says. Now, Oracle Autonomous Data Warehouse essentially handles all those things for you. It takes care of that grunt work, so you can move on to more valuable work.