Posted on May 7, 2019 by Chris Richards
Featuring an interview with Erik Benner, Mythics VP of Enterprise Transformation
When Erik Benner gets his hooks into new technology, his curiosity just will not let go. So much so that when astronomy caught his fancy, he didn’t merely buy a telescope, he ended up with a well-appointed observatory. And likewise, when he began exploring what Oracle’s most powerful engineered database system, Oracle Exadata, could do, he ended up with one in his basement.
Benner pushes new technology to the limit and is unabashed about reporting back the good and the bad. This propensity for curiosity and candor makes Benner—who owns the amorphous title of vice president of enterprise transformation at tech consulting firm Mythics—a popular speaker at tech user group events, where fellow data experts seek out his sessions.
His latest obsession is the autonomous database on Oracle Cloud. Available since 2018, Oracle Autonomous Database deploys, manages, secures, and upgrades itself with no human intervention—all tasks normally done by skilled database administrators.
Here, after more than a year of working with clients on a mix of Autonomous Database projects, are four of Benner’s suggestions: One to help you get started, two that save you money, and one gotcha to avoid.
The on-premises version of Oracle Database 19c is a single database designed to be configured and fine-tuned for your use case by experts like Benner. But Oracle Autonomous Database while using that same Oracle Database 19c running on Exadata, has been pretuned for two different kinds of work and is managed day-to-day by AI and machine learning algorithms. Those services are Oracle Autonomous Data Warehouse for analyzing tons of data all at once, and Oracle Autonomous Transaction Processing mixed workloads.
“You’ve got to pick one up front,” says Benner. “If you’re flowing lots of data into [a single place] to run analysis, then go with Autonomous Data Warehouse. If you’re not sure [what your needs will be], go with Autonomous Transaction Processing,” Benner says. “Autonomous Transaction Processing gives you a little more control, and you get a cool new feature called automatic indexing,” he explains.
And though Autonomous Transaction Processing isn’t optimized for data warehousing like Autonomous Data Warehouse, “you still have all the advantages of storage cells in the Exadata,” so you should still get good query performance.
“If I were writing an app right now, and I wanted a low-cost database option that would cloud scale, I would look at Autonomous Database,” says Benner. “Because instead of needing three DBAs for my project, I only need one.”
With Autonomous Database, “I don’t need a DBA who can build a database, I don’t need a DBA that can do all the install tasks, I don’t need a DBA that patches once a quarter. I don’t need a DBA that knows how to tune a complex [server cluster].”
You do still need that one skilled DBA, Benner says, it’s just that Oracle Autonomous Database frees him or her from routine tasks in order to do more valuable stuff. “I need a DBA who’s an application DBA; who knows how to write good SQL, who knows how to work with developers on their SQL, who knows how to structure data in a relational database,” he says. And “I need someone who knows how to get data into and out of that database. What’s the best method given our budget and our needs?”
These tasks give the DBA a more visible role. You save salary on your database project, and the DBAs you pay “get to do stuff that’s of more obvious value to the business,” Benner says.
While there are several killer use cases for Autonomous Database, one of Benner’s favorites is its true hot scaling capabilities—meaning you can scale it up or down at will. “If you’re running a production environment for a data warehouse and only do one or two huge analytics workloads a month, you can run all month on two CPUs and then when you’re going to do your big reports you can scale that thing up to 18 CPUs, or whatever you need, and take advantage of all that parallelism, knock out your analytics, then slam it back down to 2 CPUs,” he says. “That’s huge for savings and efficiency.”
Or “if I only need my dev and test environment Monday through Friday between 8 a.m. and 7 p.m., I can just shut the Autonomous Database down and stop paying for it during all those hours I’m not using it,” he says. “Now what have I done to my bill?”
One thing to watch when setting up an autonomous database is how resources are allocated, says Benner, because it’s different in Oracle Autonomous Database. “I’ve had issues with Autonomous Database where a user was getting crappy performance, and they should have had screaming-fast performance,” he says. After some digging, Benner realized the problem was very simple. An inexperienced administrator had given every database user the lowest-tier priority for access to computer resources.
Normally, a DBA would know a user’s needs and give them the resources their work requires. “But now users can pick all by themselves, and they often go cheap,” because they pay for the cloud service by the CPU hour, he says. Oracle Autonomous Data Warehouse has three tiers and Autonomous Transaction Processing has five, so if a job is running slowly, “go back and look at what tier they’re connected to. It might be too low for what they’re trying to do.”
As the tech world adjusts to autonomous databases, Benner will be among those leading the way. At a recent meeting on Oracle’s main campus, for example, Benner, who is bestowed with the company’s top user title, Ace Director, brought his suggestions for continuing to advance the autonomous database experience and shared them with Oracle executives. They were all ears.
Click Here to learn more about how Mythics can help transform your organization with Oracle Autonomous Database.