AWS
Act 1 · Room

The Monolith Room

A Tuesday, 6 days after a production outage. The CTO has a board meeting in nine days.

Gmail · Inbox6 days ago
From: Priya N., CEO <priya@acme.com>
To: Miguel R., VP Engineering
Tuesday’s outage — what’s the plan?

Miguel,

Fourth-quarter close was supposed to be quiet. Instead we spent 4 hours and 12 minutes yesterday with the orders pipeline on the floor. Again. Sales is asking how many times we can apologize to the same three customers before one of them leaves.

Finance says the GSI wants $14M over 5 years to take us off WebSphere. The board will say yes on Dec 15 unless we come back with a credible alternative. A real one — not a deck.

This cannot happen again. What’s our modernization path?

— Priya

INC-2026-1013 · SEV-1 post-mortem
Duration
4h 12m
Missed orders
$1.2M
Root cause
WebSphere thread-pool exhaustion under normal peak load
Recurrence
3rd time in 18 months
Owner
nobody — last OrdersService maintainer left in 2019
thread-pool-default: 512 / 512 · queue-depth: 8,193 · oldest-request: 3m 47s
FINAL
signature due Dec 15
Statement of Work · Hyperscale GSI
$14M
· 5 years · 40 consultants
  • • MAP-accelerated monolith decomposition engagement
  • • 38 bounded contexts, 4 modernization waves
  • • Projected first production cutover: Month 14
  • • Projected portfolio finish: May 2030
Oracle support ends Dec 31, 2027. At this cadence, 30 months late.
Oracle Database Enterprise Edition support ends Dec 31, 2027. 14 months, 03 days from today.

The GSI will take 5 years. The support contract is gone in 14 months. You need a third option.

The third optionCursor Cloud Agents + AWS
5× faster than GSI

Cursor Cloud Agents work alongside your own engineers — scanning the monolith, drafting architecture, authoring CDK, writing tests, and driving the cutover. Every gate still goes through a named human reviewer. First production cutover in 22 days, not 14 months.