Database administrators (DBAs) are under intense pressure to improve the performance of their Oracle databases. Businesses are processing more data, more transactions and larger data sets, yet they also demand faster database performance with no latency. At the same time, many organizations are embracing database consolidation to reduce costs and improve efficiencies, while still expecting improved database performance. It’s a constant theme that is unlikely to change any time soon. In fact, with the growth of big data analytics and the Internet of Things, the pressure on DBAs will become even greater.
One of the most critical steps any enterprise can take to drive database performance is to utilize all-flash storage. IT decision-makers who have deployed flash storage for databases identified “improved application performance” as the most important benefit to their organizations.1 It’s also important to note that the real performance benefits of flash only come from using an all-flash solution. With a hybrid, the flash is typically used as a cache, and when the cache gets filled the workload moves to spinning disk. At that point performance will degrade significantly, and you will once again have the storage bottleneck you were trying to avoid.
For DBAs, all-flash storage drives dramatic performance gains in a number of key areas, particularly: database IOPS, query performance, data reduction and backup. In addition, the right all-flash solution will accelerate the availability of analytics, and will reduce the amount of time required to manage the storage solution. With all-flash storage, DBAs don’t have to worry about block size, file systems, performance tuning, LUN management or RAID.
What are some of the performance gains a DBA can expect to achieve? Recent performance testing of Oracle using the FlashStack all-flash converged infrastructure platform yielded the following results:
|Metric Description||1 x //m50|
|Maximum IOPS (8KB) with latency under 1 ms across all 6 workloads||242k|
|Maximum IOPS (8KB) with latency under 1ms for 90% read, 10% update workload||336k|
|Maximum achieved ingest data load rat||5 TB/hour|
|Maximum achieved bandwidth (@510 KB)||7 GBPS|
|Average data reduction of data warehouse||3.1:1|
These results mirror what enterprises are experiencing in the real world. Based on interviews with DBAs across a number of industries, these are the types of performance gains that are being achieved with all-flash storage arrays:
- A large global insurance vendor improved transaction times by more than 15x after moving from spinning disks to an all-flash array
- Skullcandy was able to reduce the latency of their BI platform from 5000ms to sub-millisecond
- Kiabi saw a 5x improvement in Oracle batch job processing times
- Secure-24 was able to deliver highly differentiated SLA, achieving the highest levels of secure and complex application hosting services, with 3x-10x improvement in application performance
In addition, DBAs can be much more efficient by turning off Oracle RMAN compression and using an all-flash array for data reduction. For example, one organization was able to reduce the backup time of a 260 GB database to 26 minutes using Pure Storage data reduction, compared to 244 minutes using RMAN with the Oracle compression function.
The results are clear: If you are a DBA responsible for managing or deploying an Oracle database, only all-flash storage will give you the performance gains you need to meet the challenges of today’s business environments. Today, all-flash solutions are as important to the Oracle database as the Oracle database is to the business.
1“How to Make Oracle Databases Faster and More Efficient With Pure Storage,” Enterprise Strategy Group, November 2015.