An assessment that produces a defendable estimate has five parts; the items in bold are the usual effort drivers:
- Artifact inventory (from the twx exports or the repository REST): counts per type per app; heritage human services and heritage coaches (each is a rewrite as CSHS + coach views: 0.5-3 days per service), BPEL / IID modules (redesign as service flows or ACE), Java integrations (per jar: compile and test on Liberty, 0.5-2 days), file-system and OS-level scripts, custom Process Portal changes (rewrite for Workplace or drop), direct REST calls in coach views (prefix + CSRF), SQL integrations against BAW's own tables (fragile), tracking groups (fine), external services (recreate server definitions).
- Integration inventory: every endpoint with protocol, auth and TLS; MQ / JMS queues (Liberty config), SMTP, LDAP, ECM; network path from the cluster (firewall rules, egress IPs, proxies); CORS for external UIs.
- Platform and operations: OpenShift cluster sizing (production + non-production), storage classes, databases (new PostgreSQL vs migrated Db2), identity (LDAP / IdP with IAM), monitoring / logging stack, backup, CI/CD pipeline rebuild (Operations REST), runbooks and training for oc / operator.
- Data: in-flight instances per app (drain vs database migration), document stores (BPM document store → FileNet or keep), PDW history (BAI going forward), audit requirements.
- Testing: regression per app (a playback per exposed item), performance test of the top 3 processes, security test of the new front door, UAT of Workplace with business users (biggest change for them).
# quick inventory from a Process Center (classic REST): artifact types per snapshot
GET /rest/bpm/wle/v1/processApps -> apps, snapshots
GET /rest/bpm/wle/v1/snapshot/{snapshotId}?parts=validation -> deprecated / heritage artifacts appear as warnings
GET /rest/bpm/wle/v1/assets?snapshotId={snapshotId} -> managed files (jars, zips) to reviewRules of thumb from projects: an app with no heritage artifacts and REST-only integrations migrates in days (import, servers, env vars, test); each heritage human service adds days; each BPEL module adds weeks; the platform build (cluster, IAM, pipeline, monitoring) is a one-time 6-12 week effort that dominates small estates - do it with the first pilot app, then migrate the rest in waves.
References