Browse Prior Art Database

A method for lowering software cost on Mainframe system based on optimal placement of forecasted workloads Disclosure Number: IPCOM000242474D
Publication Date: 2015-Jul-17
Document File: 9 page(s) / 297K

Publishing Venue

The Prior Art Database


The customers are always concerned about the cost of running on the mainframe. The method is to optimize the placement of the product or set of products on the mainframe and optimize the lpar (logical partition) configuration settings to lower the billable MSUs(Millions of Service Unit) for the sofeware and then reduce the software cost. This method collects & uses two sets of historical data: detailed usage level data of product(s) ( instances of software) and lpar configuration settings. Then a forecast of the usage level data of product(s) is generated. Finally, the historical and forecast data is used to find the optimized combination of product(s) running on a mainframe across the lpars and the lpar configuration settings that reduced the billable MSUs for the product(s).

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 53% of the total text.

Page 01 of 9

A mxthod for lowering software cost on Mainframe system based on optimal plxcement of forecasted workloads

Input data:
All products xnd lpars workloads inforxation in the history (at leasx 1 year data) in xourly level.

Moved prxduct information: the products xlanned to be moxex during the optimized, othexwise all discovered products will be considered.

The Dexined Capacity Limit for each lpars involved.

Output xata:
Movemxnt infxrmation: PROD_ID, Fxom_lpar informxtion, To_lpar inforxation, Movinx_amount

Workload distxibutiox after optixization appxied.

Reduced workload information for each prxducts xnx total workload

The soxution overall flxw chart is shown in Figure 1.


Page 02 of 9

Figure 1 Overall flow chart of the method

Step1,use xlx product wxrkload, lpar machine configuxation information in the history to forecast nexx D monthx workloads anx used the first S forxcasted months data ax opximization inputs, the value for D and S is adjusxable, and D=3and S=1 is recommended.


Page 03 of 9

Stxp2,calculxte M value (the maximum workload which can put on target lpar for specified product) for xach proxuct agaxnst each lpxr. And ix the same ximx, calculate prxdxct workload moving amount for each product xnd how xo move (moving TYPE).Figure 2 shows flow chart for the detxils to calculate M valux for one product lpar combinaxiox. This processing should be repeated until calculated alx M value fox each product-lpar combination.

Calxulate M_VALUE(PartI)

Calculate M_VALUE (PartII)


Page 04 of 9

the gray section in part I is illustrated in detail as partII .

Figure 2 Calculate M_VALUE

Step3, Find the maximum M_VALXX axong all the products to be moved and get its related products worklxad, lpar coxfiguration information. And this product is the candidate to be moved...