You ve probably thought about the speed of your servers running your business intelligence workloads. But what about your client devices? Choosing the wrong mobile platform can leave you waiting for reports and without key functionality. How much of a difference can the choice of client device make when running IBM Cognos Enterprise reports? Prowess put four devices to the test to find out: Lenovo ThinkPad Yoga Lenovo Miix 2 Apple ipad Air Samsung Galaxy Note 10.1 Faster reports Reports on mobile devices powered by Intel architecture ran 3 7 times faster than on devices powered by ARM processors in Prowess testing. We tested typical reports run against large datasets and discovered that the Lenovo ThinkPad Yoga and the Lenovo Miix 2, powered by Intel architecture and Windows 8.1, offered clear advantages for both decision makers and data scientists. For enterprise decision makers, the ThinkPad Yoga and the Miix 2 generated reports an average of 3 7x faster than on the ipad Air or the Galaxy Note 10.1 in Prowess testing. For data scientists, accessing Cognos Enterprise from mobile devices running Windows 8.1 and powered by Intel processors delivers the capability of creating new reports beyond those already created in Cognos. Web browsers and Cognos mobile client apps available for Android and ios devices do not have this capability. Additional functionality Craft reports beyond precreated ones on the fly. Higher performance The Intel Xeon processor E7-4890 v2 and IBM DB2 with BLU Acceleration can boost performance 148x over previousgeneration hardware and software in IBM and Intel testing. 1 1
Choice of hardware can impact server performance for Cognos as well. Testing conducted by IBM and Intel reports that the latest-generation Intel Xeon processor E7-4890 v2 coupled with IBM DB2 with BLU Acceleration can increase server performance over previous-generation hardware and software. 1 Find Insights Faster and Untether Your Data Scientists with the Right Tools IBM Cognos Enterprise is a market-leading business-intelligence (BI) solution that handles some of the largest deployments evaluated in the 2014 Gartner Magic Quadrant for business intelligence and analytics platforms. 2 This made it the logical choice for testing BI apps on mobile platforms. Mobile access to Cognos Enterprise provides users with what they need to freely explore information, analyze key facts, and quickly collaborate all from wherever business might take them. IBM Cognos Unified Workspace Dashboards Reports Scorecards Ad-hoc Query Analysis & Exploration Trend & Statistical Analysis What-If Analysis IBM Cognos Data Analysis Dynamic Cubes Dynamic Query Mode Classic Query Mode OLAP Over Relational Dimensionally Modeled Relational IBM Cognos Data Sources Database Aggregates Relational Sources Large Enterprise Data Warehouse PowerCubes Modern and Legacy Sources Application Sources Third-Party OLAP Sources Figure 1. Diagram of capabilities of IBM Cognos Enterprise 2
Prowess Consulting evaluated options for mobile access to Cognos Enterprise on four different, popular mobile devices. We found that the Cognos user experience varies notably between different mobile device platforms. Of the devices we tested, we concluded that the mobile devices best suited for productive access to Cognos Enterprise were the Lenovo ThinkPad Yoga and the Lenovo Miix 2 running Windows 8.1 and powered by an Intel Core i3 processor and an Intel Atom processor, respectively. Table 1. IBM Cognos Enterprise access across four different mobile platforms (for full configuration and test environment descriptions, see Appendix A) Lenovo ThinkPad Yoga Lenovo Miix 2 (10 inch) Apple ipad Air Samsung Galaxy Note 10.1 Powered by Intel architecture and Windows 8.1, accesses IBM Cognos Enterprise from a browser interface Powered by Intel architecture and Windows 8.1, accesses Cognos Enterprise from a browser interface Running ios 7.1.1, accesses Cognos Enterprise through Cognos Mobile Running Android 4.3, accesses Cognos Enterprise through Cognos Mobile Intel Core i3-4010u processor (1.70 GHz 3 MB) Intel Atom Z3740 processor (1.33 GHz 2 MB) 64-bit ARM processor: Apple A7 chip and M7 motion coprocessor 32-bit ARM: Samsung Exynos 4412 (ARM Cortex -A9, quad-core, 1.4 GHz) 3
Depending on the data set, reports generated with IBM Cognos Mobile on mobile devices based on ios or Android can take upwards of 20 to 30 seconds to generate. For a single report this difference can be trivial, but when viewing several reports in a rapidly moving discussion referencing, the delay might weigh on productivity. 20 30 seconds to generate reports on ARM processor-based devices 3 6 seconds to generate reports on Intel processor-based devices By contrast, the ThinkPad Yoga and the Miix 2 produced reports in as little as three to six seconds in Prowess testing, even for very large data sets. In addition to speed, accessing Cognos Enterprise from a mobile device running Windows such as on the ThinkPad Yoga and the Miix 2 enables data scientists and other users to access IBM Cognos Workspace Advanced. Whereas Cognos Mobile can only render reports already constructed in Cognos, Cognos Workspace Advanced lets users adjust existing reports and dashboards or create entirely new ones. This enables testing hypotheses and following up on hunches on the go, rather than waiting to get back to a desktop workstation. Table 2. Feature comparison between IBM Cognos Workspace Advanced and IBM Cognos Mobile Feature IBM Cognos Workspace Advanced IBM Cognos Mobile Create reports from relational or dimensional data sources Add additional measures to reports Add conditional formatting to reports Add advanced calculations to reports Beyond client devices, Intel hardware can help improve server performance. The latest-generation Intel Xeon processor E7-4890 v2 and DB2 with BLU Acceleration together can boost server performance 148x over previous-generation hardware and software. 1 Intel Solid-State Drives (SSDs) and Intel 10 Gigabit Server Adapters can help lift that performance higher still while helping reduce operating costs. 4
This paper provides an overview of how enterprise decision makers and data scientists use Cognos Enterprise, and it describes in more detail what Prowess discovered in testing about the performance and capabilities of different mobile platforms in accessing Cognos reports. A Day in the Life of Cognos Enterprise Users The CEO of a national retailer is consulting a series of Cognos dashboards on her ipad Air in a quarterly planning meeting with other senior managers. The CEO uses her ipad Air for most things without a second thought, but the delay in pulling up new reports on her ipad Air upwards of 30 seconds in some cases can add up. The CEO notes on the Cognos Transaction Analysis Dashboard that Washington, D.C. is her company s largest market. Drilling down into that market, the CEO also notes on the Customer Loyalty Dashboard that Washington, D.C. also has one of the lowest participation rates by customers in her company s customerloyalty program. Figure 2. IBM Cognos Enterprise Transaction Analysis Dashboard on an Apple ipad Air 5
Sensing clues that she would like to tease out, the CEO opens the Customer Loyalty Average Spend report and notices that Washington D.C. customers enrolled in the loyalty program have spent far more than customers not enrolled in it. This contrasts sharply with Georgia, which is the company s fourthlargest market, and one with a high loyalty-program participation rate, but whose loyalty-program members spend only a little more than non-members. At this stage of inquiry, though, the CEO runs into a wall. She would like to find out reasons for the discrepancies between these markets, but she is limited to pre-set reports on her ipad Air. The CEO asks another senior manager in the meeting to dig deeper in the reports, but he encounters the same limitation on his Android tablet. The CEO knows that adding fields to the pre-set reports could get her closer to the answers she s looking for. Checking variables like the age of loyalty-program members in the D.C. and Georgia markets might offer some clues as to the differences in spending by program members in those two markets. But she is unable to change the reports on her ipad (or even on the other manager s Android) and she will have to turn to someone accessing Cognos from a Windows based computer to tailor the reports to her needs. After the meeting, the CEO tasks a data scientist with digging into the data to figure out why Washington, D.C. customers participate in the customer-loyalty program at such a low rate yet spend significantly more than customers in the same market who are not enrolled in the program. The data scientist has his Windows 8.1 tablet with him and can generate custom reports on his tablet. And because reports generate with minimal pause on his Windows tablet, he starts exploring some hypotheses with the CEO right there in her office. Adding fields to existing IBM Cognos reports is a fast way to look for new insights in data, but it is not possible on tested ARM processor-based devices. 6
Figure 3. IBM Cognos Workspace Advanced accessed from a Lenovo ThinkPad Yoga running Windows 8.1 Perhaps customer churn among loyalty-program members is to blame. The data scientist opens the Customer Loyalty Membership Rate and Contribution report and adds a field for lost customers. The new report does not reveal much, so the data scientist adds the Age Group field to the same Customer Loyalty Membership Rate and Contribution report. This new report shows that Washington, D.C. has only older customers in its loyalty program. This is another sharp contrast with Georgia, which has only younger members in the program. With some clues in hand, the data scientist agrees to prepare a more formal analysis for the CEO. However, when he departs the CEO s office, he does not need to return immediately to his desk to work on this new assignment because Cognos on his Windows client has all of the capabilities it does on his desktop. Performance Characteristics of Servers and Computers Running Latest-Generation Intel Processors with Cognos Prowess testing indicates that mobile devices running Windows 8.1 and powered by Intel processors such as the ThinkPad Yoga and the Miix 2 can generate Cognos reports faster than devices not running Windows 8.1 that are powered by ARM processors. Devices like the ThinkPad Yoga and the Miix 2 also demonstrated capabilities beyond devices powered by ARM processors to interact with Cognos. In addition, testing by IBM and Intel indicates that Intel Xeon processors can have an impact on the performance of servers running the IBM DB2 backend for Cognos. 1 7
Client-side Benefits Cognos Enterprise brings reporting, analysis, modeling, planning, and collaboration together to produce smarter decision-making and better business outcomes. Mobile access to Cognos Enterprise provides users with the ability to freely explore information, analyze key facts, quickly collaborate, and act with confidence all from wherever business might take them. To assess how the choice of mobile device impacted working with Cognos, Prowess tested the time it took each device to generate three different dashboard-style reports and one drill-down report. IBM supplied standard test data from 2013 modeled sales data for a national U.S. retailer. Reports generated on the ThinkPad Yoga and the Miix 2 consistently generated three to seven times faster than on the ipad Air or the Galaxy Note 10.1. (For full test results, see Appendix C.) Speed was not the only advantage that Prowess testing uncovered for the ThinkPad Yoga and the Miix 2 in working with Cognos reports, however. Both devices could access Cognos Workspace Advanced with its additional functionality; this workspace within Cognos enables Cognos users to create custom reports or to change existing reports to reveal new insights. Cognos Mobile clients for the ios and Android devices can generate reports already defined in Cognos Workspace Advanced but cannot actually access Cognos Workspace Advanced to create new reports or alter existing ones. This limits the questions that business managers can ask of their data on ios and Android devices. Server-side Benefits The speed benefits of latest-generation Intel processors are even more dramatic on the server side. Running DB2 with BLU Acceleration on the Intel Xeon processor E7-4890 v2 can improve performance 1.9x over the previous-generation Intel Xeon processor E7 family. Taken together, DB2 with BLU Acceleration and the latest-generation Intel Xeon processor E7 family can provide better performance than previous-generation hardware and software. 1 DB2 with BLU Acceleration takes advantage of Intel Advanced Vector Extensions (Intel AVX) on the Intel Xeon processor E7 family. Intel AVX increases the throughput of floating-point, single instruction, multiple data (SIMD) calculations like those used in highly parallelized computations. This enables DB2 to pack more data elements into the register of a single processor and then process all of those data elements with just one call. This optimizes both memory usage and processor input/output (I/O) and can dramatically increase performance for computationally intense workloads like database queries. 1 8
The Intel Xeon processor E7-4890 v2 also provides built-in, hardware-based security. This includes Intel Data Protection Technology with Intel Advanced Encryption Standard New Instructions (Intel AES-NI) and Intel Secure Key. Intel AES-NI protects business intelligence data by making data encryption faster and stronger, so that good security does not come at the price of diminished performance. 4 Intel Secure Key further strengthens database security by providing true random number generation. Data encryption has traditionally relied on pseudorandom number generation. While generally good enough, it could be compromised if malicious software could obtain the seed value used to generate the pseudorandom numbers used in encryption. Intel Secure Key counters this threat by using truly random numbers that cannot be recreated by attackers. The benefits of Intel hardware for server performance extend beyond processors. Speed matters on your storage as well. The Intel SSD DC P3700 series can provide maximum latencies of less than 500 µs 99.9 percent of the time. 3 How does this stack up against regular hard-disk drives (HDDs)? 414x 92x 30x more read IOPS more write IOPS less average latency Intel 10 Gigabit Server Adapters can also substantially increase server responsiveness. For some workloads, just replacing a 1 gigabit network adapter with one from the Intel Ethernet Converged Network Adapter X520 Series of 10 gigabit network adapters reduced latency by 80 percent. 6 Greater responsiveness from the server backend enables you to generate and change reports more quickly and arrive at insights faster. The latest-generation Intel Xeon processor E7 v2 family coupled with IBM DB2 with BLU Acceleration provides 148x better performance than previous-generation hardware and software. 1 9
Maximize Your Cognos Productivity on Intel Hardware Cognos Enterprise can put the tools for business insights in the hands of more users. Many businesses focus on speed for their server backend but might not be aware of delays on the client frontend. Our testing showed that of the devices tested, the Lenovo ThinkPad Yoga and the Lenovo Miix 2, powered by Intel processors and Windows 8.1, offered the best combination of performance and capabilities for all Cognos users. The Lenovo devices can help improve Cognos user productivity in two ways: Cognos reports generated 3 7x faster than on tested devices powered by ARM processors. Reports can complete for users in as little as 3 5 seconds, even on large data sets. Provide full access to Cognos Workspace Advanced to generate new reports and customize existing ones. Mobile devices powered by ARM processors cannot access this feature of Cognos Enterprise and can only run pre-created reports. Intel hardware can improve performance on the server side as well. The Intel Xeon processor E7-4890 v2 coupled with DB2 with BLU Acceleration can increase server performance 148x over previous-generation hardware and software. 1 Intel SSDs and Intel 10 Gigabit Server Adapters can lift that performance higher still while helping reduce operating costs. To maximize the productivity of business tools like IBM Cognos, your hardware matters on both the server and the client side. Intel processors and hardware used end-to-end in Cognos Enterprise deployments can increase productivity while reducing costs. The Lenovo ThinkPad Yoga and the Lenovo Miix 2 powered by Intel processors and running Windows 8.1 offered the best combination of performance and capabilities for all IBM Cognos users. 10
Appendix A: Hardware Details Client Devices Name & Model OS Architecture & Processor Memory Storage Apple ipad Air ios 7.1.1 ARM ; 64-bit; Apple A7 (dual-core, 1.4 GHz) 1 GB 16 GB Samsung Galaxy Note 10.1 Android 4.3 ARM; 32-bit; Samsung Exynos 4412 (ARM Cortex -A9, 3 GB 16 GB quad-core, 1.4 GHz) Lenovo Miix 2 10 Windows 8.1 Update 1 Intel Atom Z3740 processor (1.33 GHz, 2 MB) 2 GB 64 GB Lenovo ThinkPad Yoga 12.5 Windows 8.1 Update 1 4th gen Intel Core i3-4010u processor (1.70 GHz, 3 MB) 4 GB 500 GB Appendix B: Test Configuration Details Software (Client) Name Version Notes IBM Cognos Mobile 10.2.1.4.1 For testing on the Apple ipad Air IBM Cognos Mobile 10.2.1.4.1 For testing on the Samsung Galaxy Note 10.1 Internet Explorer 11 running in compatibility mode 10.2.1 For testing on the Lenovo Miix 2 and the Lenovo ThinkPad Yoga Software (Server) Name Version IBM Cognos 10.2.1.2 IBM DB2 Advanced Enterprise Server 10.5 11
Appendix C: Test Result Details Test data supplied by IBM, which simulates data for a U.S. national retailer. Report-generation timed from submission request to report generation in the client. All reports generated without pre-cached report data on the client devices. All measurements are in seconds. Test mean values measured with a confidence level of 95 percent. Run 1 Run 2 Run 3 Run 4 Run 5 Run 6 Run 7 Run 8 Run 9 Run 10 Run 11 Run 12 Run 13 Run 14 Run 15 Mean Value and Margin of Error Transaction Analysis Dashboard Lenovo ThinkPad Yoga 6.6 5.4 5.7 5.7 5.5 5.5 5.2 5.6 5.5 5.4 5.6 5.3 5.4 5.0 5.4 5.5 ± 0.2 Lenovo Miix 2 2.3 5.9 5.7 5.3 5.2 5.0 5.3 5.2 4.7 4.9 5.7 5.7 5.6 5.6 5.8 5.2 ± 0.5 Apple ipad Air 28.8 29.0 31.0 29.3 28.6 29.2 28.8 27.8 29.4 29.4 28.5 28.6 29.4 28.5 29.1 29.0 ± 0.4 Samsung Galaxy Note 10.1 50.9 32.5 34.0 34.2 32.3 34.6 35.7 34.4 33.0 33.2 34.4 36.0 35.6 35.1 34.6 35.4 ± 2.5 Drill-down Report: Washington, D.C. Lenovo ThinkPad Yoga 3.5 3.2 3.2 3.3 3.1 3.1 2.9 3.0 3.0 2.9 3.1 2.9 3.0 2.9 3.0 3.1 ± 0.1 Lenovo Miix 2 3.1 3.1 3.2 4.9 3.5 5.0 4.0 6.1 3.5 3.7 3.4 4.1 3.9 3.7 4.4 4.0 ± 0.5 Apple ipad Air 23.6 24.2 24.6 24.1 24.3 24.2 24.7 24.6 26.4 24.0 25.4 25.2 25.4 25.3 25.2 24.7 ± 0.4 Samsung Galaxy Note 10.1 27.4 28.4 27.4 28.4 27.0 29.9 27.0 27.7 27.7 27.4 25.6 28.9 29.0 28.3 28.0 27.9 ± 0.6 Customer Loyalty Dashboard Lenovo ThinkPad Yoga 5.1 5.9 5.3 5.4 5.2 5.4 5.4 4.3 5.9 5.8 5.6 5.4 5.6 5.6 5.5 5.4 ± 0.2 Lenovo Miix 2 5.3 5.3 5.4 5.8 5.4 5.2 5.3 5.5 5.4 5.5 5.1 5.3 5.3 5.2 5.2 5.3 ± 0.1 Apple ipad Air 20.3 20.5 20.2 20.3 22.5 20.5 20.2 21.2 20.7 20.9 20.5 21.5 20.8 20.9 20.2 20.7 ± 0.3 Samsung Galaxy Note 10.1 24.4 22.4 22.6 23.1 22.7 22.1 22.1 22.3 22.8 23.1 23.4 22.8 23.1 22.9 22.8 22.8 ± 0.3 Customer Churn Dashboard Lenovo ThinkPad Yoga 5.1 5.1 5.2 5.1 5.2 5.2 5.1 5.0 5.1 5.1 5.2 5.6 5.1 5.3 5.2 5.2 ± 0.1 Lenovo Miix 2 5.4 5.5 5.3 5.3 5.2 6.0 5.2 5.2 5.3 5.4 5.2 5.1 5.2 5.2 5.2 5.3 ± 0.1 Apple ipad Air 20.1 20.1 20.5 20.6 20.8 20.4 20.6 20.1 20.6 20.3 20.2 20.1 20.4 20.3 19.8 20.3 ± 0.2 Samsung Galaxy Note 10.1 22.6 23.0 22.5 22.2 22.9 23.2 22.7 22.5 22.7 22.9 22.9 22.9 24.6 22.9 23.0 22.9 ± 0.3 12
1 Test results based on an internal 10 TB database Proof of Performance and Scalability (POPS) benchmark with IBM DB2 10.1 running on a 4-socket Intel Xeon processor E7-4870 with 1,024 GB of memory, and DB2 10.5 with BLU Acceleration running on a 4-socket Intel Xeon processor E7-4890 with 1,024 GB of memory. Both systems were configured with IBM XIV Gen3 storage attached with 8 Gbps Fibre Channel (FC) via 8 GB SAN switch with 111 TB total raw space (2 TB x 12 disks/pmodule x 11 XIV modules). Performance improvement figures are cumulative of all queries in the workload. Individual results will vary depending on individual workloads, configurations, and conditions. http://www.intel.com/content/ dam/www/public/us/en/documents/white-papers/big-data-xeon-e7-v2-unlock-the-value-ibm-paper.pdf. 2 Gartner. Magic Quadrant for and Analytics Platforms. February 2014. http://www.gartner.com/technology/reprints.do?id=1-1qhksep&ct=140206&st=sb. 3 As measured by Intel: 100 GB 4K Random Writes QD=1 at 99.9% of the time across 100% span of the drive. 4 Intel. Intel Data Protection Technology with AES-NI and Secure Key. http://www.intel.com/content/www/us/en/architecture-and-technology/advanced-encryption-standard--aes-/ data-protection-aes-general-technology.html. 5 Comparison based on published data for the Seagate Cheetah 15K.7; Intel SSD DC S3700 delivers 75,000 random read IOPS compared to 181 delivered by the Seagate Cheetah 15K.7; 36,000 random write IOPS compared to 391 by the Seagate Cheetah 15K.7. 6 1.6 milliseconds average latency for a Microsoft Exchange Server 2010 workload configured with an Intel Ethernet Converged Network Adapter X520-DA2 40 gigabit network adapter versus 8.0 milliseconds average latency for the identical workload configured with Intel Ethernet Controller I350 gigabit network connection. Test results based on Microsoft Load Generator running with Exchange Server 2010 with 500, 100 MB mailboxes on a 12-core Intel Xeon processor E5-2697 v2 with 256 GM of memory. Both configurations were configured with their respective network adapters 1 TB of storage on the storage server accessed via iscsi. Configuration with the Intel Ethernet Converged Network Adapter X520-DA2 40 gigabit network adapter resulted in 36 percent greater iscsi network utilization (1,197 Mb/s versus 883 Mb/s when configured with the Intel Ethernet Controller I350 gigabit network connection). Individual results will vary depending on individual workloads, configurations, and conditions. Principled Technologies. Increase density and performance with upgrades from Intel and Microsoft. September 2013. The analysis in this document was done by Prowess Consulting and commissioned by Intel. Results have been simulated and are provided for informational purposes only. Any difference in system hardware or software design or configuration may affect actual performance. Prowess, the Prowess logo, and SmartDeploy are trademarks of Prowess Consulting, LLC. Copyright 2014 Prowess Consulting, LLC. All rights reserved. Other trademarks are the property of their respective owners. 13