GSE Nordic 2015 Conferece - CICS Opening. CICS TS V5.1 became available at the end of 2012, followed by CICS TS V5.2 in mid 2014. So what does that mean for you? Perhaps you can reduce operational complexity by hosting modern application interface logic inside the Liberty profile server within CICS? Perhaps you can use the new CICS Cloud capabilities to improve application reliability and accelerate deployments? Maybe you can improve your integration with Mobile devices? Or better control your costs though policies and consolidation? Nick and Ian open the CICS track to help you find out how CICS is reinventing mainframe application serving one again. Discover which sessions you should attend to get the most value from your attendance.
CICS data provided by John Burgess, Ian Burnett, Graham Rawson and Catherine Moxey
Performance results obtained for internal benchmarks on Hursley z13:
An internal CICS-VSAM (‘DSW’) workload using CICS TS V5.2, when run on a z13 with 2 terminal-owning regions, 4 application-owning regions and function shipping to 1 file-owning region, saw a 24% reduction in CPU per transaction in the CICS address spaces, when compared with the same workload on zEC12. This equated to a 31% improvement in the internal throughput rate (ITR).
(Upper graph) An internal CICS-DB2 workload, running CICS Transaction Server for z/OS V5.2 in a single region, and using DB2 11 and z/OS V2.1, showed a 28% reduction in CPU per transaction and a 39% increase in internal throughput rate when run in threadsafe mode on a z13, compared with running in threadsafe mode on a zEC12. The workload was run in a single LPAR with 4 CPs online, and no zIIP specialty engines.
An internal IBM workload using SSL over HTTP into CICS TS V5.2 with persistent sessions, and exploiting improved CP Assist for Cryptographic Function on z13, saw a 35% reduction in CPU per transaction in the CICS address space compared with the same workload run on a zEC12.
(Lower graph) Measurements were obtained using an internal CICS web services workload, which sent a web service request using non-persistent SSL with full handshake from a CICS region in one LPAR to a CICS region in another LPAR over TCP/IP, with a request containing ten 80-byte elements. The web service request was processed by linking over CICS MRO to a target program in a remote CICS region in the same LPAR. The target program updated each of the ten elements, and returned a response containing ten 80-byte elements. This web services workload was run on a zEC12 using CICS TS V5.2, and was compared with running the same workload using the CICS TS V5.3 open beta (developmental code) with SSL on zEC12, as well as with the same workload using CICS TS V5.3 open beta and SSL on a z13, and finally with the workload using the CICS TS V5.3 open beta on a z13 but using AT-TLS in place of SSL. This benchmark used a Crypto Express5S card on the z13 and a Crypto Express4S card on the zEC12. The Triple DES (3DES) cipher was used in all cases, but a minimum TLS encryption level of TLS 1.0 was used for the CICS TS V5.2 measurements, and a minimum TLS encryption level of TLS 1.2 was used for the CICS TS V5.3 measurements. The cumulative increase in ITR between CICS TS V5.2 with SSL and zEC12 and CICS TS V5.3 open beta with AT-TLS on z13 was 61%. Raw ITR values were 2479, 2840, 3267, and 3989 respectively.
An internal CICS client/server workload, driven using SSL over HTTP with full handshake, and using the CICS WEB send and receive API with a 32K payload, was run using CICS Transaction Server for z/OS V5.2 on z/OS V2.1. The workload showed a 22% reduction in CPU per request when run on z13 compared with zEC12. The response time as measured by RMF reporting saw a reduction of 77% on z13 compared with zEC12.