Aller au contenu

Le week-end de bascule

Aussi en 19 langues

English · Español · 日本語 · Português (Brasil) · Deutsch · Italiano · 한국어 · 简体中文 · Português (Portugal) · 繁體中文 · Bahasa Indonesia · Türkçe · Polski · Tiếng Việt · ไทย · Українська · Español (México) · Français (Canada) · Deutsch (Schweiz)

Early morning in a river valley: the road now curves onto a new concrete bridge, while the old steel bridge beside it stands empty, its road closed by a line of barriers.
Image générée par IA

Module 4 sur 5 du chapitre Migration et synchronisation, chapitre 10 du cours Data Architect. Chaque module est autonome. Previous: Automation During the Load · Next: Getting Data Out: Keeping Another System in Step ⟲

La question à laquelle répond ce module : comment basculer, et prouver que la migration a fonctionné ?

C’est vendredi soir. Les utilisateurs sont partis, l’équipe de migration est en ligne, et quelqu’un dit : « On l’a testé trois fois. Allons-y. » D’accord. Qu’allons-nous faire exactement ? Quand l’ancien système cesse-t-il de changer ? Quand a lieu l’extraction finale ? Qui désactive l’automatisation, qui lance chaque chargement ? Et si nous avons deux heures de retard à minuit, ou si 3 % des comptes échouent ? Qui décide si nous continuons ? Comment prouvons-nous que les données sont justes, et quand les utilisateurs peuvent-ils revenir ? Et si cela tourne mal : comment en sortons-nous ? La bascule, ce n’est pas « lancer la migration ». C’est une transition maîtrisée d’un état connu à un autre.

Here’s our landscape one more time. This is the weekend that amber arrow runs, and then disappears. ⟲

The arrow that runs once Steve Trucking's landscape from Checkpoint 1, in the same places, the weekend RouteDesk retires: people on the left, Salesforce with Field Service in the middle with the Retail Partner portal and the tracking page, the integration layer from chapter 9 to its right, and the systems around it: Snowflake, FleetLink, the finance ERP, photo storage, and RouteDesk, retiring, with 8 years and about 5 million deliveries. One new arrow, in amber, the migration this module zooms in on, carries what's still open from RouteDesk into Salesforce, migrated once. Underneath: one arrow runs once and then disappears; a migration is an integration with an end. {"author": "Steve Simpson", "blueprint": "1.0.1", "builder": "standard 1.54.0", "built": "2026-10-30", "id": "DGM-DA-MG-10", "rights": "\u00a9 2026 Twinshark LLC. All rights reserved.", "template": "system-landscape 0.13.0", "theme": "stevetecharc 1.4.0"} ABOVE: IDENTITY, APPEXCHANGE, WEB PEOPLE RIGHT: INTERNAL SYSTEMS BELOW: CUSTOMER CHANNELS BOTTOM RIGHT: PARTNERS AND PROVIDERS Dispatch coordinators 45, inside 9 partner DCs Delivery crews ~700 on 330 trucks Care and office staff 100: service, ops, finance Retail Partner users ~300 at 3 partners Recipients ~1 million a year Corporate IdP SSO for office staff only Retail Partner portal Tracking page INTEGRATION Snowflake status history, analytics FleetLink raw GPS stays here Finance ERP invoices partners, pays crews Photo storage 4–5 million a year, off core RouteDesk retiring: 8 years, ~5 million deliveries Phone care agents Email email-to-case SMS and email provider day-before message, with a link Redrock REST API, webhooks Keystone CSV over SFTP Brightwave EDI through a third-party network Salesforce Field Service, one US org a link, no login office staff only calls cases in manifests in status out manifests in status out manifests in status out upserts, partner key status events status events latest position only delivery counts delivery counts photos, from the crews' app day-before, with a link what's still open: migrated once Salesforce a community people identity, apps, web an internal system a customer channel the integration layer starts at the system that opens the connection; the label says what moves a shared certificate: trust, no direct call the migration we zoom in on Checkpoint 1's landscape, the weekend RouteDesk retires. One arrow runs once and then disappears: a migration is an integration with an end. © 2026 Twinshark LLC · stevetecharc.com
Cliquer pour agrandir

The weekend this arrow runs. On Monday morning, it’s gone. ⟲

But more changes that weekend than one arrow. The partners’ manifests stop landing in RouteDesk and start arriving through the integration layer. RouteDesk’s own outbound feeds, Keystone’s nightly status file and Redrock’s webhook, have to come from Salesforce instead. Coordinators route Monday’s deliveries in Field Service, not RouteDesk. And on Monday morning, 330 trucks leave the distribution centers on routes from the new system. Everything in this module is about getting to that Monday safely, and knowing what we’d do if we couldn’t. ⟲

Je veux que la bascule en production soit presque décevante. Nous avons déjà exécuté le processus, nous l’avons chronométré, nous avons vu les échecs, testé les contournements, rapproché les résultats et nous sommes entraînés à rétablir le fonctionnement normal. La production, c’est le même scénario avec de nouvelles données. Si nous inventons la migration le week-end de bascule, nous sommes déjà en difficulté.

That takes a dress rehearsal, not “we loaded accounts in the sandbox.” Beginning to end: a final-style extract, transformations, quality checks, the load sequence, automation and integration controls, reconciliation, deferred processing, the return to normal mode, all timed, at production or representative scale, in a Full sandbox. Record how long each step took, what failed, what needed manual help, what confused people, where we waited, what surprised us. Then improve the runbook. The first rehearsal teaches you how the migration works; the next one should prove you learned. And because module 2 brought the keys back, a rehearsal doesn’t have to start from nothing every time: fix one transformation, reload just the rows it touched, and run the reconciliation again. For RouteDesk, that means the whole weekend in a Full sandbox: the extract, the preparation data store, the waves of module 1, the bypass of module 3, the switch of the partners’ feeds, and the reconciliation, with a stopwatch running. ⟲

Pas « samedi matin : migrer les données ». Quelque chose de plus proche de :

ÉtapeActionResponsableDurée prévuePreuve / critère de sortie
1Confirm RouteDesk is read-only after Saturday’s last truckOperations lead10 minWritten confirmation; time of the last change recorded
2Final RouteDesk extract into the preparation data storeData team45 minCounts and control totals captured
3Confirm the bypass on the migration userSalesforce team10 minA test record saves with no status events sent
4Incremental load: Saturday’s changes, upsert by the partner keyMigration team30 minResults and new IDs back in the preparation data store
5Reconcile against the expected numbersData team and operations30 minCounts, totals and aggregate reports approved
6Switch the partners’ manifests and status feeds to the integration layerIntegration team30 minOne test manifest upserted and one status sent, per partner
……………
14Remove the bypass from the migration userSalesforce team10 minPermission set unassigned, and checked

Now we know what happens, who does it, how long we expect, and what proves each step is complete. That’s much stronger than a checklist that says “Load Work Orders.” Notice step 14. Module 3’s bypass is a permission to skip business rules, so taking it away is a runbook step with an owner, not something we hope someone remembers. ⟲

I call go-live Day 0: usually a Sunday or another off-hours day. Steve Trucking delivers six days a week, so Day 0 is the Sunday with no trucks on the road. Day +1 is after it; Day −1 and Day −7 are before it. And on any large load, I ask: do I have to wait until Day 0 to load all the data? If the load takes hours and the source keeps changing, I’m going to need an incremental anyway. So: ⟲

  • Day −7: the initial load. Load the bulk of the data a week early, kept invisible by sharing, so users and reports don’t see it. Smoke-test users (a system administrator) validate it in place. Ask the business what would be hurt by loading early; if the answer is “it would break my reports,” load it hidden. It’s much simpler to open up the sharing on Day 0 than to spend five hours loading a million records and all their children inside the window. ⟲
  • J−1 : le chargement de rafraîchissement. Reprenez ce qui a changé depuis le J−7, et protégez la ligne de départ : les utilisateurs modifient-ils encore la source, des intégrations écrivent-elles, des traitements planifiés tournent-ils ? Gelez la source, mettez-la en lecture seule, ou définissez exactement comment les changements postérieurs à l’extraction seront capturés.
  • J0 : l’incrémental. Chargez ce qui a changé depuis le J−1 (parfois un incrémental au début de la fenêtre et un dernier à la fin), rapprochez, décidez du go ou du no-go, ouvrez le partage, rétablissez le fonctionnement normal.

For RouteDesk, Day −7 is the Sunday before: the 190,000 delivered Work Orders go in, hidden by sharing, and smoke-test users check them all week. History barely changes, so most of the load is done before the weekend starts. Day −1 is Saturday night, once the last truck is back: RouteDesk goes read-only, and the refresh brings the week’s changes. Day 0 is Sunday: a small incremental, then the gates. ⟲

Day −7, Day −1, Day 0, and the rollback point Four days left to right. Day −7, the initial load: 190,000 delivered Work Orders, hidden by sharing, which smoke-test users check. A week of changes leads to Day −1, the refresh: the week's changes, with RouteDesk read-only once Saturday's last truck is back. Saturday's changes lead to Day 0, highlighted, a Sunday with no trucks out: the incremental, reconcile, go/no-go, open the sharing, and switch the partners' manifests to Salesforce. Go leads to Day +1, in coral and tagged "the rollback point": Monday, the first truck leaves on a route from Field Service, and RouteDesk stays read-only. Underneath: load early and hidden, refresh, then a small incremental on the day; rollback has a clock, and it runs out when the first truck leaves. THE CUTOVER THE POINT OF NO RETURN Day −7: the initial load 190,000 delivered Work Orders, hidden by sharing; smoke-test users check them Load early and hidden, refresh, then a small incremental on the day. Rollback has a clock: it runs out when the first truck leaves. HOW TO READ IT cutover day after this, going back means losing real work © 2026 Twinshark LLC · stevetecharc.com
Cliquer pour agrandir

Day −7: most of the data, a week early, where nobody can see it yet. ⟲

Day −7, Day −1, Day 0, and the rollback point Four days left to right. Day −7, the initial load: 190,000 delivered Work Orders, hidden by sharing, which smoke-test users check. A week of changes leads to Day −1, the refresh: the week's changes, with RouteDesk read-only once Saturday's last truck is back. Saturday's changes lead to Day 0, highlighted, a Sunday with no trucks out: the incremental, reconcile, go/no-go, open the sharing, and switch the partners' manifests to Salesforce. Go leads to Day +1, in coral and tagged "the rollback point": Monday, the first truck leaves on a route from Field Service, and RouteDesk stays read-only. Underneath: load early and hidden, refresh, then a small incremental on the day; rollback has a clock, and it runs out when the first truck leaves. THE CUTOVER THE POINT OF NO RETURN Day −7: the initial load 190,000 delivered Work Orders, hidden by sharing; smoke-test users check them Day −1: the refresh the week's changes; RouteDesk read-only once Saturday's last truck is back a week of changes Load early and hidden, refresh, then a small incremental on the day. Rollback has a clock: it runs out when the first truck leaves. HOW TO READ IT cutover day after this, going back means losing real work © 2026 Twinshark LLC · stevetecharc.com
Cliquer pour agrandir

Day −1: the week’s changes, and the source stops moving. ⟲

Day −7, Day −1, Day 0, and the rollback point Four days left to right. Day −7, the initial load: 190,000 delivered Work Orders, hidden by sharing, which smoke-test users check. A week of changes leads to Day −1, the refresh: the week's changes, with RouteDesk read-only once Saturday's last truck is back. Saturday's changes lead to Day 0, highlighted, a Sunday with no trucks out: the incremental, reconcile, go/no-go, open the sharing, and switch the partners' manifests to Salesforce. Go leads to Day +1, in coral and tagged "the rollback point": Monday, the first truck leaves on a route from Field Service, and RouteDesk stays read-only. Underneath: load early and hidden, refresh, then a small incremental on the day; rollback has a clock, and it runs out when the first truck leaves. THE CUTOVER THE POINT OF NO RETURN Day −7: the initial load 190,000 delivered Work Orders, hidden by sharing; smoke-test users check them Day −1: the refresh the week's changes; RouteDesk read-only once Saturday's last truck is back Day 0, Sunday: no trucks out the incremental, reconcile, go/no-go, open the sharing, switch the manifests to Salesforce a week of changes Saturday's changes Load early and hidden, refresh, then a small incremental on the day. Rollback has a clock: it runs out when the first truck leaves. HOW TO READ IT cutover day after this, going back means losing real work © 2026 Twinshark LLC · stevetecharc.com
Cliquer pour agrandir

Day 0: a small incremental, then the gates. ⟲

Day −7, Day −1, Day 0, and the rollback point Four days left to right. Day −7, the initial load: 190,000 delivered Work Orders, hidden by sharing, which smoke-test users check. A week of changes leads to Day −1, the refresh: the week's changes, with RouteDesk read-only once Saturday's last truck is back. Saturday's changes lead to Day 0, highlighted, a Sunday with no trucks out: the incremental, reconcile, go/no-go, open the sharing, and switch the partners' manifests to Salesforce. Go leads to Day +1, in coral and tagged "the rollback point": Monday, the first truck leaves on a route from Field Service, and RouteDesk stays read-only. Underneath: load early and hidden, refresh, then a small incremental on the day; rollback has a clock, and it runs out when the first truck leaves. {"author": "Steve Simpson", "blueprint": "1.0.0", "builder": "standard 1.48.0", "built": "2026-10-27", "id": "DGM-DA-MG-41", "rights": "\u00a9 2026 Twinshark LLC. All rights reserved.", "template": "cells 0.4.0", "theme": "stevetecharc 1.4.0"} THE CUTOVER THE POINT OF NO RETURN Day −7: the initial load 190,000 delivered Work Orders, hidden by sharing; smoke-test users check them Day −1: the refresh the week's changes; RouteDesk read-only once Saturday's last truck is back Day 0, Sunday: no trucks out the incremental, reconcile, go/no-go, open the sharing, switch the manifests to Salesforce Day +1, Monday: the first truck leaves routes from Field Service; RouteDesk stays read-only the rollback point a week of changes Saturday's changes go Load early and hidden, refresh, then a small incremental on the day. Rollback has a clock: it runs out when the first truck leaves. HOW TO READ IT cutover day after this, going back means losing real work © 2026 Twinshark LLC · stevetecharc.com
Cliquer pour agrandir

Day +1: the first truck leaves, and the rollback clock runs out. ⟲

This isn’t an unusual way to cut over; it’s the common one, under other names. Salesforce’s own guidance for moving a B2C Commerce site says to migrate as much data as you can “as close as possible prior to the actual cut-over. Ideally, only the delta from the previous day should be required to copy during the cut-over to minimize the down time” (B2C Commerce Realm Split FAQ (s'ouvre dans un nouvel onglet), Salesforce Help). Microsoft’s guidance for its own platform is a full migration “about a month before go-live, followed by weekly delta loads,” with the final delta “after freezing both source and target systems” (Data migration planning for go-live (s'ouvre dans un nouvel onglet), Microsoft Learn). And AWS’s database migration service builds the same idea in: a full load while it captures changes, then “shut down your applications, let the remaining changes flow through to the target” (Components of AWS DMS (s'ouvre dans un nouvel onglet), AWS; all three checked 2026-10-27). The timing changes with the volume: a month or a week ahead, weekly or daily deltas. The Salesforce twist is the one I add: load early hidden by sharing, so the data is in place and checked, and nobody sees it until Day 0. ⟲

Test knowledge. For a large migration with a short cutover window, load most of the data days early, hidden from users by sharing, then load only what changed (a refresh before the window, an incremental inside it). Freeze or capture changes in the source from the extract onward, and validate with test users before opening access. ⟲

For the cutover, the point is the gates around those loads. By Day −7 I don’t want to be designing the migration anymore: mappings, transformations, the load sequence, automation decisions, migration identities, reconciliation reports, the runbook, rollback criteria and known exceptions are stable, because the real loads have started. Things can still change, but every late change reduces rehearsal. At some point you stop improving the migration and start protecting the cutover. And know when the source stopped moving, or know exactly how you’ll capture what changed after your extract. ⟲

That second half is the watermark: the point through which each extract was current, so the next load knows what changed since. For RouteDesk, the watermark is the moment of each extract: everything RouteDesk changed after Sunday’s Day −7 extract is in Saturday’s refresh. That only works if RouteDesk records when a delivery last changed, so check that it does, and that every DC’s screens update it. If it doesn’t, compare snapshots: extract again and find the differences by key. Slower, but it’s a watermark you can trust. Without one, delta processing is guesswork. ⟲

Le J0, c’est alors de l’exécution. Suivez le runbook, notez les temps réels, les ID de traitement, les décomptes, les erreurs et les décisions. Si l’étape 17 échoue, ne sautez pas à l’étape 24 parce que quelqu’un pense gagner du temps : comprenez la dépendance, utilisez la procédure de reprise, mettez à jour l’état. L’équipe doit toujours pouvoir dire où nous en sommes, ce qui est terminé, ce qui a échoué, ce qui vient ensuite, et si nous sommes encore dans la fenêtre.

Bloquez l’accès des utilisateurs quand vous avez besoin de maîtriser

Section intitulée « Bloquez l’accès des utilisateurs quand vous avez besoin de maîtriser »

Personne n’aime l’indisponibilité. Mais faire comme si le système était disponible pendant que nous changeons ses fondations peut être pire, et « nous voulons zéro indisponibilité » ne doit pas devenir discrètement « nous n’avons aucune idée de la version des données qui fait autorité ». Parfois, la stratégie la plus propre, c’est de bloquer l’accès des utilisateurs, de faire la bascule, de prouver le résultat et de les faire revenir. Comment ?

  • Les méthodes fortes : geler les utilisateurs, changer leurs heures de connexion, ou restreindre les plages d’adresses IP. Ne les désactivez pas : cela casse la propriété et tout ce qui s’y rattache.
  • La méthode douce : un login flow qui laisse les gens se connecter et les dépose dans une salle d’attente : « le système est en maintenance ; revenez plus tard ». Plus aimable que des erreurs qui font croire aux utilisateurs que leur compte est cassé. Ne mettez jamais ce login flow sur les administrateurs système ; s’il casse, plus personne n’entre.
  • Gardez toujours une porte de service : un groupe d’administrateurs système (et vos utilisateurs de test de fumée) qui peuvent encore entrer.
  • Les sessions persistent. Bloquer l’accès des utilisateurs ne met pas fin aux sessions déjà ouvertes, alors prévoyez comment faire sortir ceux qui travaillent encore : un avis, une échéance, ou, avec un développement spécifique, un moyen de les déconnecter.

At Steve Trucking it’s Sunday, so the people to plan for are the care agents answering phones, the coordinators getting ready for Monday, and the 300 Retail Partner users who might log in to the portal to check a claim. Login flows are assigned by profile (Custom Login Flows (s'ouvre dans un nouvel onglet), Salesforce Help, checked 2026-10-27), so the partners’ profile can get the waiting room (“your deliveries are moving to a new system; back Monday at 6 a.m.”) while the smoke-test users and administrators come and go. ⟲

N’improvisez pas cela la semaine d’avant. Décidez de ce que vivront les utilisateurs, et dites-le-leur.

Test knowledge. To keep users out during a cutover, freeze their accounts, restrict login hours or IP ranges, or use a login flow (assigned by profile) as a waiting room. Don’t deactivate users: it breaks ownership and what’s attached to it. Keep administrators and test users able to log in. ⟲

À 2 h du matin, après douze heures, ce n’est pas le moment d’inventer les critères de go/no-go. Définissez-les avant la bascule : objets critiques rapprochés, taux d’échec sous un seuil approuvé, aucun défaut de données critique non résolu, relations obligatoires validées, intégrations critiques vérifiées, automatisation rétablie et testée, performance acceptable, validation métier faite, assez de temps restant pour les étapes en aval. Sinon, « go » devient « on est allés trop loin pour s’arrêter ». Ce n’est pas de l’architecture. C’est le biais des coûts irrécupérables.

“We have a rollback plan.” Good. Until when? We migrate, open Salesforce, and for six hours users create cases, update accounts, enter opportunities, change contact details, while integrations send data elsewhere. Can we restore Friday night? Not without losing legitimate work. Rollback gets harder the longer the new system runs, so define the rollback point: before it, we return to the previous state with the tested procedure; after it, recovery needs a different strategy. Rollback has a clock, and the runbook should show it. For Steve Trucking the clock is easy to read: until Monday’s first truck leaves on a Field Service route, we can still send the coordinators back to RouteDesk, which is why it stays read-only and untouched until then. After that, every delivery made, photo taken and status sent lives only in Salesforce, and going back would lose real work. From that moment, recovery means fixing forward. ⟲

Le retour arrière, ce n’est pas non plus simplement supprimer ce que vous avez chargé. L’automatisation a peut-être créé des enfants, les intégrations ont peut-être envoyé des enregistrements ailleurs, les utilisateurs ont peut-être modifié des enregistrements migrés, des fichiers ont peut-être été joints, des événements publiés, des enregistrements existants mis à jour. Ces effets distribués sont la raison pour laquelle les contrôles d’automatisation et d’intégration du module 3 relèvent de l’architecture de la bascule.

And what makes you roll back? In my experience it’s rarely the loader. It’s unknown data: you find out something key was missed, or a data feed was running that nobody accounted for. The best rollback plan is not needing it: know all of your key data, and every feed that touches it, before cutover. RouteDesk has two that are easy to forget, because they go out, not in: the nightly status file Keystone picks up, and the webhook that tells Redrock what happened. Retire RouteDesk without moving them, and on Monday night Keystone hears nothing. Nobody loaded the wrong data; a feed just stopped. ⟲

Après le chargement, je ne demande pas « l’API a-t-elle dit succès ? ». Je veux quatre types de preuve, et je les utilise tous les quatre : décomptes, totaux, vérifications par sondage et rapports agrégés.

Les décomptes. Combien d’enregistrements devraient exister, et combien existent ? Comptes source sélectionnés : 500 000. Doublons consolidés : 20 000. Exclus selon une politique approuvée : 5 000. Attendus : 475 000. Cible : 475 000. Bien, mais les décomptes seuls ne sont pas une preuve ; vous pouvez avoir le bon nombre d’enregistrements et les mauvaises données.

Les totaux. Là où les chiffres comptent, comparez des totaux de contrôle : montant des factures, valeur des commandes, solde ouvert, quantité. Les deux systèmes ont 100 000 factures, mais l’ancien totalise 412 800 000 $ et Salesforce 398 100 000 $. Stop. Les totaux révèlent ce que les décomptes manquent.

Spot checks. Look at real records, not just easy ones: a simple record and a complex one, a large customer and a small one, one with many children, one that was transformed, one involved in dedupe, one with unusual history. Trace each from source to target: fields, relationships, ownership, history, External IDs. Averages hide serious individual failures. ⟲

Les rapports agrégés. Comparez la forme : comptes par État, commandes par année, requêtes client par statut, clients par segment, chiffre d’affaires par mois. Si la source dit 82 000 clients en Arizona et Salesforce 8 200, nous avons un indice : un code d’État qui n’a pas été mappé, une année exclue, un statut mal traduit, un fichier sauté. Le nombre de lignes peut avoir l’air juste alors que la distribution raconte l’histoire.

For RouteDesk, each proof has its check, and each catches something the others miss. ⟲

Four ways to prove the data Four equal panels, four ways to prove the data. Counts, how many: at RouteDesk, 196,400 expected, less the quarantine, against those loaded; catches records lost or loaded twice; misses the right number of wrong records. Totals, how much: line items per partner, against what finance invoiced; catches values changed on the way; misses which records are wrong. Spot checks, real records end to end: a revision, an install, a merged duplicate, a claim; catches a wrong field or relationship on a real record; misses problems in records nobody picked. Aggregate reports, the shape: deliveries by DC by week, beside RouteDesk's; catches a DC or a status code that didn't map; misses one bad record in a good total. Underneath: no single proof is enough; set every expected number before the load, then use all four. {"author": "Steve Simpson", "blueprint": "1.0.0", "builder": "standard 1.48.0", "built": "2026-10-27", "id": "DGM-DA-MG-42", "rights": "\u00a9 2026 Twinshark LLC. All rights reserved.", "template": "comparison 0.4.0", "theme": "stevetecharc 1.4.0"} Counts how many AT ROUTEDESK 196,400 expected, less quarantine, against loaded CATCHES Records lost or loaded twice MISSES The right number of wrong records Totals how much AT ROUTEDESK Line items per partner, against what finance invoiced CATCHES Values changed on the way MISSES Which records are wrong Spot checks real records, end to end AT ROUTEDESK A revision, an install, a merged duplicate, a claim CATCHES A wrong field or relationship on a real record MISSES Problems in records nobody picked Aggregate reports the shape AT ROUTEDESK Deliveries by DC by week, beside RouteDesk's CATCHES A DC or a status code that didn't map MISSES One bad record in a good total No single proof is enough. Set every expected number before the load, then use all four. © 2026 Twinshark LLC · stevetecharc.com
Cliquer pour agrandir

Each proof catches what the others miss. Use all four. ⟲

The aggregate report is the one I’d watch first. Remember module 1’s status codes: nine DCs, nine spellings. If DC 7’s deliveries by week come out 40 percent short against RouteDesk’s, the count for the whole load might still look close, but one DC’s DONE/INST didn’t map. ⟲

Test knowledge. Prove a migration with more than “no errors”: record counts against expected numbers set before the load, control totals on amounts that matter, spot checks of real records traced end to end, and aggregate reports that compare the shape of the data. Business owners validate it too. ⟲

Deux règles les accompagnent. D’abord, fixez les chiffres attendus avant le chargement (module 1) : décompte source, exclusions approuvées, doublons consolidés, enregistrements en quarantaine, cible attendue. Puis comparez et expliquez l’écart. Un chiffre attendu inventé après avoir vu le résultat, ce n’est pas du rapprochement ; c’est de la rationalisation. Ensuite, zéro erreur peut quand même être faux. Mappez Billing State vers Shipping State et chaque ligne se charge sans aucune erreur d’API : exécution technique parfaite, architecture fausse. C’est pour cela que le rapprochement métier compte, pas seulement les journaux techniques.

L’équipe technique peut prouver les décomptes, les clés, les relations et le succès du chargement. Les utilisateurs métier reconnaissent ce que nous ne verrons pas : « ce ne sont pas nos clients actifs », « ce statut n’est plus utilisé depuis 2022 », « ces comptes sont rattachés à la mauvaise région », « ce client devrait avoir trois contrats actifs ». La validation métier est donc une étape explicite de la mise en production, avec des personnes nommées, des rapports nommés, des enregistrements nommés et une approbation nommée, pas « jetez un œil quand vous aurez le temps ».

Why not move everything on Day 0? The Number of Zeros. If the full migration takes 18 hours and the business tolerates 4 hours of downtime, we have a problem, and moving data earlier solves it. History goes in before cutover while the legacy system stays authoritative; the window only has to handle what changed, what couldn’t be preloaded, final relationships, validation and reconciliation. Instead of moving eight years of history during the outage, we’re closing a gap. ⟲

But preloaded data needs protection. If production data lands on Day −7, can users see it, edit it, can automation react, can integrations publish it, can agents act on it? The answer should be: not yet. So we keep it hidden by sharing, owned and controlled, with automation and integrations held back (module 3). A preload isn’t “put the records in early.” It’s put them in early without accidentally making the new system authoritative early. ⟲

Before cutover, the legacy system is authoritative and Salesforce holds a staged, increasingly current copy. During cutover we close the gap. At the agreed moment, authority moves, and afterward Salesforce is authoritative. For Steve Trucking that moment is Day 0: from then on, a delivery’s truth lives in Salesforce, and RouteDesk is a read-only record of the past. That’s where chapters 6, 9 and 10 meet: identity, authority, movement and cutover aren’t separate problems; they’re views of the same data lifecycle. ⟲

So don’t let two systems both think they’re master. If the legacy system stays writable after cutover, a coordinator changes a delivery window in Salesforce, another changes it in the legacy system, and both systems think they own it. Last write wins? Salesforce wins? Human review? If that’s unanswered, we haven’t finished a migration; we’ve created two writable copies. Chapter 9’s rule: technical permission isn’t architectural authority. ⟲

Not every migration needs ongoing synchronization. If we’re replacing the legacy system completely, Day 0 comes, the final deltas load, reconciliation passes, users move, the old system goes read-only and is eventually archived or retired. Don’t build a permanent integration because “migrations usually have integrations.” If the old system is dead, stop feeding it. That’s RouteDesk: on Monday it’s read-only, then it’s retired, and nothing syncs with it. The synchronization Steve Trucking does need is a different arrow, Salesforce to Snowflake, designed in chapter 9 and covered in module 5. ⟲

Migration ends; synchronization goes on In the migration region, a defined start and end, three days left to right. Day −7, the baseline: the gap between the systems is a week of changes. Day −1, the refresh: the gap is a day. Day 0, highlighted, the final delta: the gap is hours, and authority moves to Salesforce. After cutover, two cards. RouteDesk, read-only, then retired, tagged "the end": the migration ends, and there's nothing to sync. And Salesforce to Snowflake, in green, tagged "goes on": status changes every day, from the cutover watermark. Underneath: migration has a defined start and end; synchronization is an ongoing process to keep two systems consistent. {"author": "Steve Simpson", "blueprint": "1.0.0", "builder": "standard 1.48.0", "built": "2026-10-27", "id": "DGM-DA-MG-52", "rights": "\u00a9 2026 Twinshark LLC. All rights reserved.", "template": "cells 0.4.0", "theme": "stevetecharc 1.4.0"} MIGRATION: A DEFINED START AND END AFTER CUTOVER Day −7: the baseline the gap: a week of changes Day −1: the refresh the gap: a day Day 0: the final delta the gap: hours; authority moves to Salesforce RouteDesk: read-only, then retired the migration ends; nothing to sync the end Salesforce to Snowflake status changes, every day, from the cutover watermark goes on a week of changes a day's changes migration complete synchronization Migration has a defined start and end. Synchronization is an ongoing process to keep two systems consistent. HOW TO READ IT authority moves synchronization: no end date © 2026 Twinshark LLC · stevetecharc.com
Cliquer pour agrandir

The gap closes, authority moves, and the migration ends. The Snowflake feed goes on. ⟲

The opposite happens too. A transition that lasts months, users active in both systems, business units moving at different times: synchronization may have to start before the migration is finished. Now the migration is still moving the historical baseline while synchronization keeps live records consistent, and that’s a much harder architecture. You need to know which process owns which changes, or the load and the sync overwrite each other. So define the boundary visibly: before this point we’re building the target state; after it we’re maintaining consistency. It might be a timestamp, a cutover event, a business-unit transition, an object-by-object switch, a watermark. If nobody can tell me whether a record belongs to the migration process or the synchronization process, we have overlapping ownership, and that’s where records get missed or processed twice. ⟲

Restore normal operation deliberately: automation, scheduled jobs, integrations, event consumers, agents, user access, in an order that matters. For Steve Trucking: take the bypass away from the migration user, switch the Change Data Capture subscriber back on from the cutover point (so Snowflake gets Monday’s changes, not the weekend’s load), run module 3’s one backfill, then let the coordinators in. Remember module 3’s catch-up storm: when an integration comes back, what does it think happened while it was off? When users come back, is every critical service actually ready? The return to normal is part of the cutover. ⟲

Puis nous sommes en production, mais pas terminés. Pendant une période de stabilisation (appelez-la hypercare), surveillez les erreurs de migration, les échecs d’intégration, l’automatisation inattendue, les nouveaux doublons, les relations manquantes, les problèmes de performance, les problèmes de données signalés par les utilisateurs et les dérives du rapprochement. Le risque de migration ne disparaît pas au moment où les utilisateurs se connectent.

Then clean up: temporary migration users and permissions, bypass settings, staging data, temporary scripts, migration-only configuration, files holding sensitive source data, temporary integrations, runbook access. Don’t leave migration scaffolding attached to production because “we might need it someday.” Temporary architecture needs an exit plan too. Don’t forget the preparation data store. After RouteDesk, it holds eight years of Recipients’ names, addresses and phone numbers in its raw tables. Keep what the retention policy says to keep, as evidence, then delete the rest, on a date someone owns. ⟲

Et gardez les preuves : l’extraction finale, la version de la transformation, les fichiers de chargement, les résultats des traitements, les fichiers d’erreurs, les rapports de rapprochement, les approbations, le runbook avec les temps réels, la décision de go/no-go et les exceptions. Trois semaines plus tard, quelqu’un demandera si un enregistrement était faux avant la migration ou à cause d’elle. Vous devez pouvoir répondre. Faire confiance, mais vérifier ne s’arrête pas quand le chargement est terminé.

L’IA peut aider à surveiller la bascule. Le go/no-go ne lui appartient pas.

Section intitulée « L’IA peut aider à surveiller la bascule. Le go/no-go ne lui appartient pas. »

L’IA est utile sous la pression d’une bascule : résumer les échecs, classer les erreurs, comparer des rapports, repérer des anomalies, suivre le runbook, faire ressortir les écarts par rapport à la répétition. Mais le go/no-go combine des preuves techniques, l’impact métier, le temps restant, la position par rapport au retour arrière et la tolérance au risque. L’IA peut faire ressortir les preuves. Ce sont les personnes responsables qui prennent la décision.

Au-delà de l’examen : ce que l’IA change. L’IA peut accélérer la surveillance de la bascule et le rapprochement, en particulier en trouvant des schémas dans de gros volumes d’erreurs et de données de validation. Gardez une responsabilité humaine explicite sur le go/no-go, le retour arrière et l’acceptation d’un risque de migration non résolu.

  • The rehearsal depth, the Day −7 / Day −1 / Day 0 load plan and the rollback point (what loads early, hidden by sharing) and the runbook that makes production cutover repeatable. ⟲
  • Comment l’accès des utilisateurs est bloqué puis rouvert, avec une porte de service et des utilisateurs de test de fumée, et comment l’automatisation, les intégrations et les agents sont maîtrisés pendant ce temps.
  • Les critères de go/no-go, la procédure de retour arrière et le point au-delà duquel le retour arrière cesse d’être la reprise sûre, ainsi que les données et les flux inconnus qu’il faut trouver d’abord.
  • Comment la migration est prouvée par des décomptes, des totaux, des vérifications par sondage et des rapports agrégés, par rapport à des valeurs attendues fixées avant le chargement.
  • Quelle validation métier, quelles preuves et quelle stabilisation après la mise en production sont exigées avant que la migration soit terminée.

Les scripts ont tourné : pas assez. L’API a renvoyé un succès : pas assez. Les nombres de lignes concordaient : toujours pas assez. Je veux des décomptes, des totaux, des vérifications par sondage et des rapports agrégés, une validation métier, le système revenu à la normale, savoir si nous sommes encore dans la fenêtre de retour arrière, et la preuve de ce qui s’est exactement passé. La bascule en production doit être ennuyeuse, parce que nous avons fait le gros du travail de réflexion avant le vendredi soir. Faire confiance, mais vérifier. Ensuite, laissez revenir les utilisateurs.