IBM z Systems mainframe and right-time insights: The tale beyond the tape
Today, the IT world is abuzz with the introduction of the next generation of the mainframe, IBM z Systems. The new flagship of the mainframe world, the IBM z13, comes with a pretty impressive stat sheet:- Forty percent more total capacity
- Three times increase in memory
- Two times increase in cache
- Single-instruction multiple-data (SIMD)
- Simultaneous multithreading (SMT)
This is all great stuff, but what does it mean to you and the challenges you face every day? Specifically, can this new box improve your ability to capture new opportunities at the point of impact?
You betcha!
Let’s start with “mega-memory.” There’s a lot of focus on in-memory computing (IMC) these days, especially for performing analytics against data warehouses. IBM DB2 with BLU Acceleration—the next generation database technology for in-memory computing—is now available for Linux on z Systems. The 10 TB of available mainframe memory provides a whole lot of room for innovative DB2 BLU solutions on z Systems. Turn on BLU and get answers from your Linux data orders of magnitude faster and with very little effort!
Although DB2 for z/OS is not formally classified as an IMC database, many clients today already execute 90 percent or more of their DB2 transactions in memory. The availability of three times more memory, a new cache design, improved input/output bandwidth and compression can help serve up more data faster for analysis.
While I’m on the topic of DB2 for z/OS, I’d like to point out that the IBM Competitive Project Office just concluded a study comparing z13 with attached IBM DB2 Analytics Accelerator to a comparably configured competitive system, and found that the IBM configuration delivered 17 times the performance and 13 times the price performance of the competition for a representative enterprise-class workload (161,166 concurrent reports of widely varying complexity driven by 80 concurrent users). The Accelerator can already deliver amazing results to clients (see my blog post on the topic); z13 raises the bar even further.
The DB2 Analytics Accelerator plays a significant role in the z Systems analytics strategy, and this appliance can evolve to support new and increasingly innovative use cases, such as in-database transformation and advanced predictive analytics. These enhancements can enable faster and more sophisticated reporting and allow for deep analytics application integration. For example, IBM is now doing preliminary new-technology studies with clients showing a 10 times improvement in modeling speed. This can give these clients a competitive edge by allowing them to build and refresh sophisticated predictive models faster and with less effort, which in turn allows them to serve their customers more effectively.
Linux on z Systems is already well known for its ability to consolidate workloads from distributed systems. Cognos is an ideal sort of workload to take advantage of Linux virtualization (see my blog post on how IBM uses z Systems to consolidate all of its Cognos needs onto just two servers). With SMT enabled, clients who run Cognos on Linux under z/VM could see up to a 60 percent increase in throughput per core with the new z13. An already strong consolidation story is now downright irresistible!
Predictive modeling is critical for many high-value business analysis functions, such as portfolio optimization, vehicle routing and delivery scheduling, network capacity planning and financial forecasting. This sort of complex mathematical modeling is a processor-intensive activity. Being able to produce rich, accurate models—within time windows that are usually heavily constrained—requires fast and specialized processing. The prior generation of mainframe, the IBM z Enterprise EC12, has been a proven performer for handling these types of applications. The z13 takes this performance to a whole new level.
Compared to the zEC12, the z13 is showing significant improvements in the performance of two key mathematical libraries: z/OS MASS (Mathematical Acceleration Subsystem) demonstrates up to a 2.9 times improvement on the new machine, and up to 6.8 times improvement for functions that have been improved through exploitation of the new SIMD capability. Similarly, z/OS ATLAS (Automatically Tuned Linear Algebra Software) improvements have been measured at up to 44 percent overall, and 80 percent for an SIMD-enabled function.
Imagine being able to understand what action will delight your customer and improve your business results—an upgrade offer, free shipping, a cross-sell opportunity—in real time, at the moment of engagement; with z13, this is the new possible!
The premiere resource optimization product from IBM,IBM CPLEX Optimizer on z/OS, has been modified to use the new z13 SIMD capability and has demonstrated up to 33 percent overall improvement in throughput and 75 percent improvement with SIMD enabled.
Imagine being able to adjust operations with agility so that you can react to changing conditions—reroute transportation patterns, change the order of trade settlements, redeploy finite resources—while always maximizing business results. With z13, this is the new possible!
IBM takes the mainframe’s role in analytics very seriously, and we’ve been doing so for quite a few machine generations! These are just some examples of how today’s new generation of mainframes further solidify the role of z Systems as a premiere “system of insights.” Why not take a look at how z13 can help you capture new opportunities at the point of impact?
z13: imagine the possibilities!
Paul DiMarzio has 30+ years experience with IBM focused on bringing new and emerging technologies to the mainframe. He is currently responsible for developing and executing IBM’s worldwide z Systems big data and analytics portfolio marketing strategy. You can reach Paul on Twitter: @PaulD360.
