Zero-Downtime Custom-Code Cleanup: ATC Automation + Blue/Green Cut-Over

TL;DR (90 w)

Moving to S/4 fails when millions of Z-lines block ATC checks and Go-Live windows are counted in minutes. We show how a Micro-GCC squad automated ABAP Test Cockpit (ATC) pipelines, fixed 77 % of findings robotically, and executed a blue/green cut-over that kept the plant floor read-only for just 11 minutes. Terraform, Jenkinsfile, and a one-line Python diff-tool are included.

Why Z-Code Haunts S/4 Projects (170 w)

Every ECC customer has 0.5–3 M lines of Z-objects. In theory you:

1. Run ATC S/4HANA readiness check.
   
2. Fix critical findings.
   
3. Transport green code to S/4.
   

Reality:

• 10 000+ ATC findings, mostly low-risk, overwhelm teams.
  
• Manual fixes collide with month-end close.
  
• 6–24 h outage windows for SUM (Software Update Manager) are unacceptable for 24 × 7 plants.
  

Solution: automate 80 % of fixes and switch traffic using blue/green so users barely notice.

ATC Automation Pipeline (290 w)

Metrics (1.2 M Z-LOC project):

ATC report dropped from 10 230 » 1 855 critical issues—sprint scope, not death march.

Blue/Green Cut-Over Strategy (260 w)

1. Blue Environment – ECC running legacy Z-code.
   
2. Green Environment – S/4HANA 2023 stack + cleaned transports.
   
3. Replicate Data – SLT real-time to HANA target; freeze window only for delta.
   
4. Read-Only Switch – At T-11 min put ECC in read-only; users see banner.
   
5. SUM Downtime Phase – Import final delta transport to Green.
   
6. Route Traffic – F5 BIG-IP toggles DNS; UI5 front-end hits Green.
   
7. Back-Out Plan – If KPI smoke fails (< 2 min), F5 reverts to Blue.
   

Measured downtime (plant MES tests): 11 min 18 s—within agreed 15-min SLA.

Code Snippet – Auto-Fix Script (Python 45 lines)

python

CopyEdit

import re, pathlib, sys

OLD_NEW = {

    r”(?i)cl_gui_alv_grid”: “cl_salv_table”,

    r”\bws_upload\b”:       “gui_upload”,

}

def fix_file(path):

    txt = pathlib.Path(path).read_text(encoding=”latin1″)

    for pat, repl in OLD_NEW.items():

        txt = re.sub(pat, repl, txt)

    pathlib.Path(path).write_text(txt, encoding=”latin1″)

if __name__ == “__main__”:

    for f in pathlib.Path(sys.argv[1]).rglob(“*.abap”):

        fix_file(f)

Runs in Jenkins before ATC; cut findings 18 %.

Monitoring & Rollback (130 w)

• ATC delta – Grafana panel shows open critical findings per week.
  
• Cut-Over KPIs – login success, RF scanner latency, batch-job backlog.
  
• Alert Rule – if RF latency > 500 ms for 2 min → auto-revert F5 pool to Blue.
  

Smoke test failure at T+4 min triggers unattended rollback; plant ops never notice.

Cost & Time Benchmarks (120 w)

ROI: $96 k dev cost saved; Go-Live four months earlier.

Common Pitfalls & Fixes (150 w)

Take-Home Checklist (60 w)

1. Export Z-code to Git via abapGit.
   
2. Run static auto-fix script.
   
3. Batch ATC in Jenkins; waive low-risk.
   
4. Merge & transport to Green stack.
   
5. Execute blue/green switch with 15-min read-only window.
   
6. Monitor KPIs; rollback if red in first 5 min.