Enhancements Throughout

To leverage the trust economy, the IBM z14* system features many enhancements from the core out to help accelerate business. It features up to 170 configurable cores—up to 35 percent more total capacity compared to the IBM z13*—in a single footprint.

Pervasive encryption is enabled with capabilities including an increase of crypto instructions of the processor by almost 4 to 7x. CryptoExpress 6S cards have a performance boost by a factor of 1.5 to 2x. The new data set encryption enables clients to encrypt data in bulk to leverage integrated crypto hardware, and leverage the encryption capabilities without making any application changes and no impact to business service-level agreements.

The memory footprint increases by a factor of 3x over the previous generation; the z14 system will have up to 32 TB of memory for the system and up to 16 TB per LPAR. It features fault-tolerant redundant array of independent memory designed to support availability. New asynchronous I/O IBM zHyperLink technology is a direct connect short distance link designed to deliver low latency connectivity between z14 and FICON* storage systems, enabling a 10x latency reduction. It enables clients to cut application response time up to 50 percent without requiring application changes.

IBM has also made investments for the z14 to take advantage of modern development. Java* is being optimized with 50 new instructions for Java performance, scalability and security on the new system. Pause-less garbage collection is now available so the application doesn’t have to wait for the memory to be dumped during the normal Java pause-less garbage collection process, resulting in a 10x improvement in the consistency of response time for applications. This is critical in analytic-heavy workloads like machine learning that are built on Java applications. Read more about what can be accomplished with Java on z14 in “A Step Up,” page 51. Node.js also has new hardware instructions and runs 2.5x better when co-located, with a 20 percent better performance compared to the z13.

Traditional workloads like COBOL also see improvements on z14. New instructions in the single instruction, multiple data facility offers a boost for traditional workloads using decimal operations (i.e., COBOL V6.2 and PL/I V5.2) and above that provided by the faster processor. IBM Automatic Binary Optimizer for z/OS*, which enables performance optimization without recompiling a COBOL application, can see a performance boost of up to 35 percent. The C/C++ complier for compute intensive application can also see performance gains.

—V.D.

Advertisement

Mainframe News Sign Up Today! Past News Letters