All upgrades need a fallback plan to previous version. This is incase the upgrade results in unexpected system degradation or errors. Even with through testing in a testbed environment, when the upgrade hits the real-world operational system, that is exposed to factors that may not be able to be simulated in testbed, the unexpected always occurs. A quick way to fallback to the system state prior to upgrade must be built into the upgrade procedure. What also must be considered is if the upgrade will be beneficial to the user immediately. A upgrade that results in extensive retraining is not beneficial to the organization. Having to take time away from actively supporting operations/production for training is a sign of a bad upgrade.

Chaque système doit bénéficier de son jumeau ou jumelle #twins, et même que dans un scénario idéal, un troisième lien ou support en parallèle, est de mise. Dans tous les cas, lors de mise à jour, si vous êtes dans une grande entreprise, ou même une PME, si vous ne maîtrisez pas le "code", alors vous n'êtes pas en contrôle total. Donc, une sauvegarde et un système en parallèle (voir même un 3ieme) sera nécessaire. Advenant le moindre problème, vous serez en mesure de réinitialiser à la case départ rapidement, et ce, en minimisant les coùts, ou les arrêts temporaires pour l'entreprise.

Upgrade on parallel run! Or test the upgrades on a test environment first before pushing the updates to the production environment

You're facing system upgrades. How can you anticipate and neutralize potential risks effectively? ----------------- Have a solid understanding of all processes ongoing to avoid outages, and engage to coordinate? It's a dead end or loaded question, unless there is clarification. Solid a/b failover testing? Coordinate with the org? Build processes that the org is accustomed to...? Too vague, pass on giving a real answer. Ken