Cisco Live 2015 is taking place next week in San Diego and we’re excited to share a new solution and present in the educational training zone! Cisco and SanDisk® have been working on a high performance Oracle Database 12c solution, powered by the latest UCS B420 M4 blade server and Fusion ioMemory PX600 flash application accelerators.
With Oracle database 11g support switching to extended support, Oracle customers considering upgrading to Oracle Database 12c should particularly look forward to this solution, as it delivers easy to deploy, high performance results, alongside excellent user scalability. There are a number of customers already taking advantage of Cisco enterprise-class UCS blade and rack servers with Fusion ioMemory PCI-e flash storage to accelerate their Oracle deployments, and we’ll be sharing a customer case study in one of our presentations.
High Oracle database performance with plenty of compute headroom
Swingbench is a commonly used tool for Oracle database testing to assess business critical Oracle Database 12c transactional workloads. With this tool, the order entry transactional schema is created and loaded with a 1.7 TB data set, and the Oracle System Global Area (SGA) is configured with 98GB (5% of dataset in memory). The transactional workload is executed with database users scaling from 50 to 400 concurrent users.
Our test results, and the database and system performance metrics were captured for analysis and reporting – see the results below:
I’d like to highlight three key solution benefits based on the test results:
- This solution delivers excellent user scalability performance results, scaling from 50 to 400 concurrent users, while delivering close to an amazing 1 million transactions per minute! (see Figure 1)
- The consistent performance, delivering 100K sustained IOPS during transaction workload with consistent 0.2 milli second latency for entire test duration (see Figure 2, ) enable business critical applications (24 X 7) to support linear user scalability performance, without compromising on latency or IOPS to deliver peak system performance.
- A customary bottleneck slowing down database applications in traditional storage are I/O issues from storage subsystem leading to wasted CPU cycles (as shown in Figure 3). This solution not only improves database application CPU utilization by reducing I/O wait cycles significantly, it also leaves up to 50% head room of compute capacity (cores) for additional database workload as shown in figure 4. This configuration setup helps to optimize a better return on investment for software licenses with increased CPU utilization.
These test results are from default configuration setup and performance results can be further increased with additional tuning and optimization.
Join Us at the Cisco Live Technical Education and Training Zone
At Cisco Live, make sure to attend the following technical sessions in the educational program. This will be a great opportunity to get a first-hand view of the Cisco and SanDisk partnership solutions and learn about the benefits you can take advantage of for your database environment.
Cisco UCS Servers and Fusion ioMemory Flash storage: A Production Deployed Customer Case Study Date: Tuesday, June 9, 2015 Time: 3PM Speaker: Jamen Koos, product manager Cisco UCS PCI-e storage, GPUs
Dramatically Accelerate Applications on Your Servers with SanDisk’s Fusion PCIe Flash Date: Wednesday, June 10, 2015 Time: 12.50 PM Speaker: Maryam Sanglaji, SanDisk product marketing manager
Cisco UCS B420 M4 enterprise-class blade server with Fusion ioMemory flash storage is an excellent choice in accelerating Oracle database 12c, and provides a platform for high-performance database consolidation. Unlike the complexities associated with traditional SAN setup, this solution is easy to deploy, cost effective and provides linear user scalability to meet business growth demands.
Stay tuned for the joint solution brief “Deploy a high performance database solution“ by Tushar Patel Prinicipal Engineer, Cisco UCS Performance and Solutions, Data center group and myself. We’ll share the complete details and benefits of this solution shortly on this blog. Stay tuned!
Follow @SanDiskDataCtr and join the Cisco Live conversation on Twitter with #CLUS and #CiscoUCS.
Prasad brings experience in building reference architecture and best practices for relational databases and NoSQL databases.