|
purebill.com Stephen Jones writing on billing and application migration |
![]() |
| . | Home | . | About | . | Archive | . | Links | . | Billing | . | Reference | . | Subscribe | . | Search | . | . |
Column - 21 April 2004 You have just 52 weekends in a year!SummaryIf your organisation performs core system changes away from business days, you can reserve a complete year's weekends with little trouble. So when planning, allowance must be made if data migrations (or other changes...) impact your application. Actions
A year's plan for an applicationFor your application, a year's weekends could be reserved (generically) in the following manner:
Total: 23 to 27 weekend reservations This leaves 25 to 29 weekends (i.e. the remainder of 52 weeks) available. Your reservations may vary if you release software more often, or if your application is impacted by the software releases of other upstream or downstream applications. As well, these reservations do not include situations where end-of-month weekends are disqualified for reasons of financial risk. As always, your mileage will vary based on your specific circumstances. In the example above, a maximum of 12 to 14 weekends per year are left available for data migrations. This allows each data migration to have a 'backup' weekend available if the primary weekend cannot fully complete. If these backup weekends are not available, the 'sunk' cost / effort / benefit expended for staff training, communications and other interdependent activities could be lost or sharply diminished. Staffing ImplicationsWhilst the operational staff supporting your application do not change, the staff performing the 'reserved weekend' activities will. For example, the software development team is likely to be different to the data migration team. This situation can impact your support staff due to:
Impact of external applicationsIn addition to change in your own application, upstream and downstream applications may require reservations that impact you as well. These impacts may include changes to your interfaces (or the manner in which they are called), the population of existing fields, or new transactions across an existing interface. Like your own software releases, their releases may not be successful and require backout and re-release. Development LeadtimesThe development leadtime determines when a planned release's scope must be locked in (or at least firmly agreed). This is related to an application's development capacity (number of days or dollars available), the size of the ideas developed, and the speed an idea can be turned into an implemented change. Due to their complexity, larger ideas can take longer to go through the development process. If there is sufficient development capacity, a percentage of a release can be reserved for 'unexpected' development - e.g. responses to a competitor, remediation of a large production defect. Where the idea's scope impacts multiple applications, your own plus those upstream or downstream of yours, it is important to identify which application has the longest leadtime This allows the idea's impacts to be confirmed for that application first, and progressively for the remainder with shorter leadtimes. All impacted applications can then develop to a common implementation point. Applying symmetry, applications upstream or downstream may impact your application as well. They may have leadtimes shorter or longer than your application's, and may need you to reserve weekends in your planning period. It is important to understand whether your application's leadtimes are generally longer than your interfacing applications (and so you require them to notify you early in their release planning), or are shorter (requiring you to inform others of specific date commitments). Central point of review - distributed control by applicationThe staff (including management) that support an application understand its operation the best. They understand the impacts of a change and will be the ones to fix any problems. For this reason, the responsibility for planning and administering the changes that will impact an application must be controlled by its management and senior operational staff. Avoiding too narrow a focus and providing an external review of planned change (i.e. bringing light into dark places), a centralised and knowledgeable review of planned changes is desirable. This review can be broken into two parts. The initial review of changes that are months away - this is the review where 'politics' is most likely to be on display (as competing interests attempt to get their plans in place) and decisions will be required between competing ideas. Closer to their implementation, additional review points will validate each change's readiness to progress, and will approve (or reject) that progression. Publication of the release plan is desirable and provides one 'point of truth' where committed reservations (and their details) can be easily discovered. Information captured for a 'weekend' eventThe following list includes details desirable for each weekend reservation (or for any other planned change):
Tags: Migration, Conversion [ Share with others ] Post this page to a social bookmarking site:
Other 'purebill' columnsPrevious column: Processing dimensions influence throughput and ability to scale Next column: VoIP - Not good enough... yet. All previous purebill columns can be found in the archive section. Recent Updates
Sign up to receive a brief text email when a new purebill column is published. JUMP TO TOP
|
. |
| Comments welcome: stephenjones(at)purebill.com | Stephen Jones © 2004-2010 - Copyright and reprint rules | Sitemap | . |