Oracle 19c RAC Performance Root Cause Fix

Title: Oracle 19c RAC DB Performance Diagnosis: Root Cause, Implemented Permanent Fix and Proof of Results (Fixed Price, 7 Days)

The problem :
Some of our batch jobs on Oracle 19c RAC run slowly unless we apply a suitable SQL Profile to the queries in the job. Once the profile is applied, the job runs fast, but other jobs then need the same treatment. We want the expert to determine why this happens and fix it permanently, so no per-job profiles are needed. We are not prescribing any approach. You drive the investigation and deliver the permanent solution.

Deliverable :
A written root cause analysis, plus a permanent fix that removes the need to apply a SQL Profile to affected queries in the jobs. The expert carries out the investigation, identifies the issue and implements the fix. Our team can provide guidance on reproducing the issue, but all analysis and implementation is done by the expert.

Acceptance criteria (the final report is accepted when it contains all six)

1)The root cause, with evidence from our AWR/ASH data and execution plans.
2)An explanation of why the SQL Profile fixes our jobs, meaning what the profile corrects that the database gets wrong on its own.
3)A systemic fix: the changes that correct the underlying cause so jobs run well without per-job profiles. Each change comes with exact commands or settings and an explanation of why it applies across jobs, not just one.
4)Evidence the fix is general: the same root cause appears in at least two different jobs, and which other jobs are likely affected.
5)Results and rollback: before/after run times on a job that currently needs a profile (test copy and production), what result means success, and tested steps to roll back.
6)A policy for the leftovers: when a SQL Profile or SQL Plan Baseline is still appropriate after the fix, and how to tell a one-off statement problem from a systemic one.

Commercial terms :

Fixed price: SGD 800. No hourly billing. Part amount is funded in escrow at the start.
1)Milestone 1, SGD 80 (10%): Initial hypotheses and work plan, due by day 2. Released when the expert delivers a short written note containing:
3 to 5 ranked hypotheses for why our queries need a SQL Profile, each tied to specific evidence in our AWR/ASH reports and plans. For each, state whether it would explain why many jobs need profiles or only one. Generic causes with no reference to our data do not count.
For each hypothesis, the test the expert will run to confirm or rule it out, and what result would confirm it.
What the expert needs from us, with any missing data requested now.
A day-by-day plan to reach the final report by day 7.
2)Milestone 2, SGD 120 (15%): Root cause findings with evidence, due by day 4. Released when the expert shows which hypotheses were confirmed and which were ruled out, with supporting numbers from our data, and explains what differs between the fast and slow runs.
3)Milestone 3, SGD 600 (75%): Fix implemented and proven, due by day 7. Released when:
The fix is implemented on our test copy, and a job that currently needs a profile, run without it, finishes within 15% of its profile-tuned runtime on three consecutive runs.
The same fix is applied to production during a supervised session, and the first scheduled production run confirms the result.
The final report meets all six acceptance criteria.
Runtimes come from the job scheduler or batch logs, with screenshots or exports included in the report.
If a result is not achieved, the expert gets one correction round. If it still fails, Milestone 3 is not payable.
If we have not scheduled the production session within 5 business days of the test-copy proof, or the verifying production run does not occur within 10 days of rollout through no fault of the expert, Milestone 3 is released.
One round of clarifications and corrections is included. Further work is a separate contract.
Stop clause: If Milestone 1 is not delivered by day 2, or an evidence-backed root cause is not delivered by day 7, either party may end the contract. Payment is limited to milestones already completed.
Our commitments: On day 1 we provide AWR/ASH reports and plans from a slow run and a fast run for two or three different jobs that need profiles, plus SQL IDs and run times, and access to the test copy. We also provide one named contact who responds within 1 business day, and we answer the expert’s data requests promptly. If we cause a delay, the deadline moves by the same amount.

Access and confidentiality :
Test copy with representative data volumes, via a limited-privilege account. Production access is only during a supervised screen-share session, using an account we create, revoked immediately afterwards. A verified backup or restore point exists before any change. All changes are scripted and reversible, and the expert provides the scripts with a plain-English summary of each before the production session. Sensitive names will be masked. NDA required.

Environment :
Oracle 19c RAC on Red Hat Linux, 3 nodes, non-Exadata, batch-heavy workload, approx. 7 to 8 TB database.

Required experience
8+ years of Oracle performance tuning, proven RAC work on 19c, expert reading of AWR, ASH, SQL Monitor and DBMS_XPLAN output, and experience implementing changes safely, with rollback, in client environments.

Pre-requisite :

1) Describe one similar RAC/batch tuning case you personally solved (root cause and how long it took).
2) Share a link to verifiable work or reviews (LinkedIn, blog, Oracle ACE profile, talks, or Upwork reviews mentioning Oracle tuning).
3) Include your availability (India time overlap preferred).
4)Shortlisted candidates will have a 20-minute live technical call where they will read a sample execution plan and AWR excerpt.
5) Generic or copy-pasted proposals will be rejected.