Huawei Success Story: RPMI

"The industry as a whole is looking to maximize the performance of its existing infrastructure. By leveraging new technologies (such as SSD storage), it is possible to increase the performance of legacy applications by supplementing the existing storage and server environment. "

Brian Scorer, Technology Services Manager, RPMI

The applications (running on the existing SAN infrastructure) also consumed significant staff resources to support the business. For example, a daily output of administration reports is required, which run on Windows 2008 servers. These reports are run using 'batch-scripts' by reading over 2 million records in a sequential database, which historically took up to 8 hours to complete. The reports are a daily requirement, so RPMI operated a night-shift team between 10PM and 6AM. Another key output for RPMI consists of "Valuation Reports". These reports took up to 46 hours to complete, which took other systems completely off-line for days at a time, often over the weekend.

Any solution that could reduce these staff-intensive processes would deliver a strong return on investment (ROI).

The ICT infrastructure consists of over 16 blade servers (and 150+ virtualized servers) located in two physically separate data centers. There is a direct mirror between locations. RPMI also utilizes the services of an off-site Disaster Recovery center to continuously create an exact replica of all storage data.

The original storage infrastructure (still in use) consists of a SAN network using SAS and SATA drives and is managed/virtualized using FalconStor's Network Storage Server (NSS) solution. The NSS provides increased flxibility in SAN management and a centralized storage administration. This infrastructure was upgraded in 2011, therefore RPMI had to look for technological improvements (not wholesale replacement of the existing SAN) to realize performance improvements. At the Disaster Recovery center, The existing SAN (100% SATA) was running over-capacity and needed replacement.

Solution

RPMI faced a dilemma. It needed to increase the performance of its legacy applications. However, a full upgrade of their SAN was not considered feasible since the system was recently installed in 2011. Instead, RPMI chose to increase the performance of the existing SAN by supplementing with an additional SSD storage solution.

In order to find a solution, RPMI turned to long-time advisor C>Ways (http://www.c-ways.co.uk/), a Huawei partner. Established in 1995, C>Ways provides a wide-range of consultancy, network infrastructure, storage and virtualization services. Their relationship with RPMI includes the design of a resilient data center solution and a major server/storage virtualization project.

Following an extensive analysis of their legacy systems, the existing storage network, the business needs, and the new performance goals, C>Ways and RPMI determined the following requirements:

• Increase overall system performance: RPMI required an increase in its storage network performance, specifially an increase in IOPS (input/output operations per second) and a reduction in latency.
• Reduce work-time for critical tasks and processes: The performance of RPMI's legacy and proprietary applications was identified as requiring increases in operating speed, for example requiring a nightshift to process. The successful storage solution would greatly reduce the time for these tasks.
• Support for Virtualization: RPMI's SAN is managed/virtualized using FalconStor's Network Storage Server (NSS) solution. The SSD upgrade needed to improve the performance of specifi FalconStor processes. • Reliability: The highest levels of uptime must be delivered to meet the demanding needs of a fiancial organization.
• Excellent TCO: RPMI operates in a competitive environment, so Total Cost of Ownership (TCO) is important. TCO includes the purchase cost, long-term maintenance costs, data center space, and power consumption.
• Future Proofig: RPMI required a SAN which would support virtual desktops in the future.
• Continuity: RPMI needed to maintain its existing investment in SATA storage, so any performance improvement solution must integrate with the existing SAN infrastructure.

"RPMI needed a new method to get more performance out of their existing SAN. We knew this meant SSD, and Huawei was able to deliver the cost and performance required. RPMI gained more capacity, superior performance with the lowest TCO in the market."

Nick Jackson, Managing Director, C>Ways

Huawei and C>Ways worked closely to design and implement a SSD storage system that fulfills these requirements. The result was the implementation of 2 Huawei Dorado2100 G2 SSD Storage systems, each containing 24 SSD disks with a capacity of 100GB per disk (with an overall capacity of 4.8TB). One of each Dorado has been installed at the RPMI data centers in Darlington and London.

RPMI has also installed a Huawei OceanStor2600T at its remote Disaster Recovery center, completely replacing the existing SAN there. The OceanStor 2600T, which contains 95% SAS and 5% SSD disks, is also virtualized using FalconStor.

End-User Benefits

Foremost, the Dorado has dramatically improved performance of legacy applications, which allowed RPMI to eliminate a night shift and to use the staff on other projects, such as improving overall IT services:

"We have been able to re-distribute the overnight shift into other areas of the business, allowing the Group IT team to concentrate on providing better services during core business hours and release resources back into the enterprise. The performance gains were dramatic and immediate, with clear business benefis."

Brian Scorer, Technology Services Manager, RPMI

RPMI was able to measure and report specifi benefis resulting from the upgrade to Dorado. According to Brian Scorer of RPMI: "The move onto SSD has been a huge success and has provided the following benefis:

• Reduced time to run the main RPS batch system. The batch now completes before the night shift person has started eliminating the need for night shift cover to run the RPS batch.
• Disk performance when running "Valuation Reports" a large report took around 46 hours to complete, the same report took 12 hours to complete. The knock on effect is that valuation runs no longer need the RPS system to be off-line for whole weekends.
• Case for unit's process used to take 20 minutes to start this now takes less than 1 minute. The total process that on average took around 90 minutes is now taking around 20 minutes.

The Dorado solution also accelerated the performance of the existing SAN network by improving the performance of the FalconStor Network Storage Server (NSS) virtualization solution. Specifially, the SSDs allow RPMI to use various FalconStor disk processes which accelerate I/O performance.

"The SSDs have provided a huge boost in performance when compared to normal disks. The two disk processes "HotZone" and "SafeCache" are utilized by our FalconStor units to provide a performance boost our original back end SAN units."

Brian Scorer, Technology Services Manager, RPMI

"The use of virtualization with the Dorado SSD solution allows us to offer high-performance simply and effectively to servers that need it. This ensures that the SSD was used for the right applications or processes. Once this happened, the performance was staggering."

Nick Jackson, Managing Director, C>Ways

Beyond performance, RPMI and C>Ways also appreciated the comprehensive service offered by Huawei. They were present in all aspects of the implementation and made sure that both the C>Ways and RPMI staff were fully trained on the new technology. Furthermore, Huawei went 'above and beyond' in terms of ensuring that the solution was appropriate for the UK market. For example, the fan speed on the Dorado was set to operate in a much warmer environment compared to the typical UK data center. Therefore, the fan was running unnecessarily, wasting power and creating excess noise.

"Huawei quickly recognized the issue, identified the problem and fied it within 2 weeks. Other vendors would probably force us to go through second and third line support before even acknowledging it."

Brian Scorer, Technology Services Manager, RPMI

Finally, the TCO of the Huawei solution was clearly superior to any other solutions RPMI considered. While SSD offers great performance improvements, the costs remain high compared to traditional disk arrays. Therefore, RPMI had to carefully choose a solutions provider that could meet their performance requirements while still delivering a very competitive TCO. Specifially, they enjoyed the following cost savings over the life of the solution:

• Initial outlay: The Huawei solution's initial network was very competitive in terms of capital expenditures. Furthermore, the SSD solution allowed RPMI to increase performance while utilizing the existing SAN infrastructure.
• Power Consumption: RPMI realized up to a 90% power savings compared to a traditional disk array through the low-power consumption of only 320W per single enclosure. Also signifiant, by utilizing SSD disks (and the SafeCache process from FalconStor), the usage of the (power-hungry) SAS disks in the existing storage system dropped approximately 50%.
• Data Center Space: At only 2U in size, the Dorado takes up signifiantly less space (to reach 100,000 IOPS would require 48U of traditional disk array).
• More effective use of staff resources: The Huawei solution has already reduced the time staff takes to run critical processes and reports. RPMI expects this to increase as applications are updated to run more effectively with the SSD disks. This will free up more IT staff to contribute to the core business of the company.

"From the perspective of Technology Services the use of SSD has provided increased throughput and decreased the wait time for disk access, increased the number of disk IOPS from the Hitachi SAN, and future proof the storage allowing RPMI to utilize Virtual desktops in the future."

Brian Scorer, Technology Services Manager, RPMI