Today's guest post comes from Bob Thome, Vice President of Product Management at Oracle.
Wandering the hallways earlier this year at Collaborate 2018, I had the opportunity to speak with quite a few Oracle customers about their thoughts on Oracle Exadata X7-2. At the risk of oversimplifying their thoughts, I think I can break them into two camps—those who see strong value in extreme performance, and those who aren’t so sure they need it. Everyone believed Exadata performance is impressive, with 350GB/sec throughput and almost 6 million read IOPS, but what about workloads that only need a fraction of these numbers. Do they see any benefit?
The answer, as always, is “it depends.” However, more often than not, even databases with relatively low performance requirements will benefit from Exadata’s performance. More specifically, running an application on an extreme performance platform can reduce the overall resources required for the application, resulting in the need for fewer cores, and fewer Oracle database licenses. In addition, Exadata’s overall performance and capacity make consolidation of a greater number of databases a real possibility, and consolidation is a numbers game. The more databases you can consolidate, the greater the efficiencies, the fewer operating systems and servers to purchase, support, host, and manage, and, with Oracle Multitenant, the fewer instances to manage. In the example below, various workloads are consolidated into a single server, sharing spare capacity. Since workloads peak at different times over the year, consolidating these various workloads saves substantial CPU resources when compared to each running on independent hardware.
Empirical tests of various workloads have shown that customers can consolidate four times more databases on the same hardware because of the features of the smart Exadata software. Pushing work down to the storage cells dramatically affects the amount of data the database servers need to process. The ability of the storage servers to scan immense quantities of data in parallel eliminates the need for many storage indexes, which also eliminates the need to maintain those indexes. Hybrid Columnar Compression not only reduces by a factor of ten the amount of data that must be stored, it makes analytics on that data more efficient as filtering can be done without reading the entire row. Exadata’s high-speed InfiniBand communications and generous flash cache ensures the database spends little time waiting on IO, even when storing data on cost-effective high-capacity, but relatively slow, HDDs. Unlike a standard server and storage, Exadata is database aware. It knows what work is latency sensitive and can prioritize resources for that work to optimize performance. All these performance features reduce the overall time the database server, and connected applications, spend idling, waiting for data.
Swiss Post highlighted the benefits of their consolidating onto Exadata at a recent OOW. They took 480 databases running on 100 physical and 30 virtual servers, and migrated them to three Exadata half-racks. They also took 96 SAP databases running on 47 physical and 15 virtual servers, and moved them to two Exadata quarter-racks. They realized large benefits in patching and overall management of the system. They increased their management efficiency from 85 databases per DBA, to 125 databases per DBA, eliminated dozens of servers, and reduced their overall costs. Perhaps one of the biggest benefits was the single point of accountability–they went from 10 hardware and software vendors to one, eliminating the multiple teams involved in troubleshooting, allowing them to more rapidly resolve issues.
The impact of extreme performance, however, can be much larger than enhancing consolidation. As summarized by Jonathan Walsh, Head of BI & DW at Morrisons, Plc.,
“With query times dropping from minutes to seconds, Exadata has changed the way people work.”
Hundreds of customers have adopted Exadata, not to speed up existing workloads, but to enable new workloads. Analytic tasks that were impossible or impractical in the past, now becoming routine. Businesses are using Exadata to analyze data in real-time, to make better decision based on huge stockpiles of data.
A large financial institution in the US moved to Exadata, and found it enabled them to perform more effective fraud detection on transactions. Deeper more sophisticated fraud analysis enabled by Exadata's performance allowed them to more reliably in real-time detect fraudulent transactions while reducing false alarms.
High performance can also help customers better meet their performance and availability SLAs. Overall database performance can easily be adversely affected, maybe due to a workload spike, a change in workload patterns, or even a change in the application. Exadata’s extreme performance ensures such events will be absorbed by the system, reducing the likelihood that response times or throughput metrics will violate the standards you’ve negotiated with your end users.
So, to me, it's obvious that performance does matter. Think about what you can do with an extreme performance database systems. Look at your opportunities for consolidation and think about what you can save. More importantly, talk to your end users and application developers. Find out what they could do with better performance and stricter SLAs. What strategic value can they derive from their IT systems if they could process data in a fraction of the time it takes today?
This is blog 2 in a series of blog posts celebrating the 10th anniversary of the introduction of Oracle Exadata. Our next post on Oracle Exadata will also focus on Performance, but will explore why only an Engineered System can deliver this level of performance.
Bob Thome is a Vice President at Oracle responsible for product management for Database Engineered Systems and Cloud Services, including Exadata, Exadata Cloud Service, Exadata Cloud at Customer, RAC on OCI-C, VM DB (RAC and SI) on OCI, and Oracle Database Appliance. He has over 30 years of experience working in the Information Technology industry. With experience in both hardware and software companies, he has managed databases, clusters, systems, and support services. He has been at Oracle for 20 years, where he has been responsible for high availability, information integration, clustering, and storage management technologies for the database. For the past several years, he has directed product management for Oracle Database Engineered Systems and related database cloud technologies, including Oracle Exadata, Oracle Exadata Cloud Service, Oracle Exadata Cloud at Customer, Oracle Database Appliance, and Oracle Database Cloud Service.