MIPS Equivalent Sizing for IBM CICS Cobol Applications Migrated to Microsoft Azure
Published Jul 04 2019 06:07 AM 4,338 Views
Senior Member

One of the key drivers behind Mainframe Modernization to the Microsoft Azure Cloud Platform is lower cost on a more agile and  flexible environment.  An environment that can run your existing Mainframe workloads using the powerful set of tools for system management, DevOps and rich analytics. 

 

Recently, Asysco, a Microsoft Mainframe Modernization Partner, performed the zRef Benchmark in Azure IaaS using their AMT GO Mainframe Modernization solution and achieved amazing results that exceeded a similar set of tests Asysco performed with larger, on premises hardware.  If you are not familiar with the zRef Benchmark, it was created by HPE (Hewlett Packard Enterprise) from the TPC-E transaction set to be representative of a typical Mainframe workload with an online (CICS) and batch component.  The zRef Benchmark test were originally deployed and implemented on an IBM Z System Mainframe to establish a baseline for MIPS.  Leveraging Azure M Series VMs, Asysco was able to deliver over 18,000 MIPS on a single system and over 28,000 on a multi-server system!  This sets a new highwater mark for Mainframe workloads in a Cloud Platform and is comparable to an IBM mid-level z14 processor at a fraction of the cost.

 

The following whitepaper, will provide a detailed overview and in depth explanation of how Asysco achieved this amazing result.  Some of the key points to consider are:

 

  • Azure virtual machine Infrastructure as a Service (IaaS) is highly scalable and cost effective.
  • The database used was SQL Server Enterprise and is a high performance alternative to DB2 for relational data management.
  • Azure Storage is a high performance and high availability alternative for Mainframe DASD data storage.
  • Both high capacity and high availability are provided though load balanced parallel instance deployment of AMT GO, i.e., IBM Parallel Sysplex-like capabilities.
  • Batch workloads can be run concurrent with online (CICS COBOL) while still meeting online end-user Service Level Agreement (SLA) commitments.

The bottom line is that for the same amount of workload, the same or better levels of service can be achieved on infrastructure with a much lower cost of ownership by leveraging the Asysco AMT GO solution on the Microsoft Azure Cloud.

 

1 Comment
Version history
Last update:
‎Jul 04 2019 06:07 AM
Updated by: