Midrange Unix Migration Process
Original Publication Date: 2004-Jul-16
Included in the Prior Art Database: 2004-Jul-16
Problem: How do you scope, plan, schedule and deploy Software Currency, (commonly referred to contractually as "N-1") across a large number of disparate, legacy systems ? Solution: The Midrange Unix Migration Process (MUMP) attempts to solve several of the more common problems with strategic outsourced service delivery in a large Midrange account. The main advantage of this methodology is that it fuses two independent support views of the major objects in the migration. 1. the Developer's / Business Unit's focus - that is the application as a whole (as it supports the business functions). 2. the Systems Operation's focus - this is each individual server on which the application runs. The two views are fused in the methodology by having the migration analysed by application but scheduled via host. This wholistic view means that both sides (Developer/Business Unit and Operations) issues are recognised and solved rather that being in conflict. Features: * The process analyses the scheduling requirements of the migration from an application-centric point of view. This benefits for the Application Developer Teams. It then schedules each of the changes that are required by host. This benefits the Operations Teams. * Tools with embedded "intelligence" allow for varying experience levels. * The process has been designed with the expectation that the Operations and Developer Departments have differing abilities, procedures, motivations and goals. * The process takes into consideration the need to explore any Application (or host) to Application (or host) dependencies. This effectively means that the impacts of migration and/or exemptions are better understood to the community as a whole. * Although being initially designed for total infrastructure refresh, the process could be adapted for partial infrastructure refresh (not all S/W is replaced), new account transition (hosts are brought up to a standard where they may be effectively managed via IBM GSA), consolidation (host are relocated from disparate locations into a data centre) and rationalisation (applications being serviced by multiple low capacity hosts are relocated onto fewer, highter capacity hosts).
Midrange Unix Migration Process
Introduction The key to resolving the problem is to understand how to organise and deliver a technology refresh cycle to the managed community of servers.
There are several elements that must be kept in focus:
Performing a change on a host will nearly always cause some kind of end-to-end impact. Very few
servers may be modified in isolation without some sort of impact. Applications Development teams tend to have a single focus (their Applications).
Operational teams tend to have a wider focus (all of the servers that they are responsinble to maintain
in a run state). Customers tend to have a systemic view of their applications. They want to understand the impacts to
the Application, not to the underlying hosts. Operational or technical support groups tend to have a lower-level focus. They think in terms of hosts
, not Applications.
Process summary (Key to dataflow diagrams)
The MUMP is split up into three distinct phases:
Migration Analysis - This sub process identifies the nature of the migration. It gathers profiling data on
each Application at the supporting host level. It then provides strategies to assist with the migration of each Application by focusing on the changes that need to be made to each host .
Dataflow diagram for Migration Analysis:
Migration Scheduling - This sub process examines to interdependencies between all of the hosts in the
community. It provides an end-to-end view of al...