Developments in commodity hardware, cheap storage and open source have motivated IBM mainframe pricing to drop...
Mainframes host a huge quantity of historical and operational data that's too expensive to move. But that inertia doesn't apply to newer workloads, such as processing mobile transactions.
IBM has made several attempts to change the economics of big iron versus distributed servers: Specialty engines make traditional distributed applications cheaper on the mainframe, but programming barriers and the "generosity factor" (an assigned percentage of time that a given workload can run on an SE) frustrate users. System Z New Application License Charge (zNALC) has a lot of conditions for mainframe administrators to meet, such as identifying specific types of new qualified applications to run on a separate logical partition (LPAR), following a specific naming convention.
Then came IBM's millions of service units from mobile device-based workloads.
CPU usage identified from mobile workloads must be directly attributable to executed transactions. Users cannot claim the discount based on distributed system metrics or estimates. However, neither the WMP nor supporting documentation defines "mobile device" strictly; it may mean tablets and smart phones, but not laptops. Browser-based transactions from mobile devices are included in the discount, even though they originate from a so-called traditional Web interface.
As for IBM software, eligible programs include currently supported releases of DB2 database software, CICS application servers, IMS transactional database management, MQ messaging middleware and Websphere Application Server. Users must have at least one zEC12 processor on the mainframe to participate.
The human workload
IBM's previous attempts to bring new workloads onto the mainframe required shifts in configuration, and proved difficult to manage and to obtain. MWP is a nice change of pace.
The basis for the MWP discount must be directly gathered from the accounting data of the transaction processors. Flagging mobile transactions might involve a server application adding identifying information into messages it sends to the mainframe. The information could be a flag in the input message, a unique transaction ID or special logon ID. IBM requires a "yes" or "no" indicator, but no further specifics, such as the mobile operating system or device type.
Once the mainframe receives the message, it requires some mechanism to propagate the mobile workload information into the back-end reporting system. Unique transaction or logon IDs are the simplest method, because they only require changes to back-end reporting systems. If the mobile flag is part of the input messages, CICS-based mainframe shops can easily configure user areas in their CICS monitoring facility records and relocate the flag into a performance record.
To participate in the discount to IBM mainframe pricing, you must submit your plan for identifying mobile transactions to IBM for approval. Upon approval, you sign a contract addendum titled, "The System Z AWLC and AEWLC Addendum for Mobile Workload Pricing" (number Z126-6300), then download the Mobile Workload Reporting Tool (MWRT).
MWRT replaces the Sub Capacity Reporting Tool (SCRT) that many IBM shops already use. MWRT takes in the System Management Facility type 70 and 89 records, and the user supplies a comma separated value (CSV) file containing the number of mobile CPU seconds per LPAR per hour.
Like SCRT, MWRT computes the peak four-hour rolling average for each LPAR in the shop. Then it discounts the mobile CPU seconds -- supplied by the CSV file -- by 60% and applies it to the entire software stack, not just to the transaction processor that ran the transaction.
Thus, if 30% of your online CPU comes from mobile devices, you could see an 18% discount in monthly license charges.
IBM has care and feeding details for MWRT in the MWRT User's Guide.
About the author:
Robert Crawford spent 29 years as a systems programmer, covering CICS technical support, Virtual Storage Access Method, IBM DB2, IBM IMS and other mainframe products. He programmed in Assembler, Rexx, C, C++, PL/1 and COBOL. Crawford is currently an operations architect based in south Texas, establishing mainframe strategy for a large insurance company.