Non-SAP Erajaya
Tiering approach & funnel methodology
| Aspect | Explanation |
|---|---|
| Challenge | 26 accounts with diverse workloads; highly uneven cost contribution (80/20 pattern). Examining all accounts equally is inefficient. |
| Core Strategy | Cost-based tiering. Effort follows money: most expensive accounts receive deepest analysis. |
| Funnel Flow | Broad first (baseline all accounts) → narrow to expensive accounts (deep dive) → cross-account analysis → roadmap. |
| Tier 1 - Deep Dive | Accounts contributing approximately 80% of cost (typically 3-6 accounts). Resource-level analysis + workload owner workshops. |
| Tier 2 - Standard | Medium accounts (approximately 15%). Pattern-based quick wins: gp2→gp3, unattached EBS/EIP, snapshot, log retention. |
| Tier 3 - Light Scan | Small/dormant accounts (approximately 5%). Check active/dormant status; consolidation or decommission candidates. |
| Commitment (RI/SP) | Calculated at payer level, NOT per account, for optimal coverage. |
| Shared-cost | Network/Shared/SSO costs driven by Non-SAP workloads analyzed as dependencies. |
| Constraints | Read-only / non-intrusive. No resource resize/delete/modify. Recommendation execution by Erajaya team. |
| Tiering Principle | Dynamic assignment from actual data, not hypothesis. 'Small' accounts can be promoted if unexpectedly expensive. |
26 accounts in Non-SAP OU - final tier determined from actual cost data
| No | Account | Account ID | Workload Type | Tier |
|---|---|---|---|---|
| 1 | Erajaya - AMS | 911167898578 | Utility/special | T1? |
| 2 | Erajaya - Barito | 654654329067 | Data/Kafka | T1? |
| 3 | Erajaya - Databricks | 142083399789 | Data/analytics | T1? |
| 4 | Erajaya - LLM Orchestrator | 225599959737 | AI/LLM | T1? |
| 5 | Erajaya - Tableau | 339712824349 | BI/analytics | T1? |
| 6 | Erajaya - 3PL | 381492132095 | Business application | T2 |
| 7 | Erajaya - DRPPS | 975050254224 | Business application | T2 |
| 8 | Erajaya - Inbound | 571600847201 | Business application | T2 |
| 9 | Erajaya - JDA | 654654481495 | Business application | T2 |
| 10 | Erajaya - POS | 654654392631 | Business application (retail) | T2 |
| 11 | Erajaya - POSOFF | 851725414623 | Business application (retail) | T2 |
| 12 | Erajaya - ROP | 891376924230 | Business application | T2 |
| 13 | Erajaya - SCM | 886436926356 | Business application | T2 |
| No | Account | Account ID | Workload Type | Tier |
|---|---|---|---|---|
| 14 | Erajaya - SDP | 891377283119 | Business application | T2 |
| 15 | Erajaya - TMS | 975050245402 | Business application | T2 |
| 16 | Erajaya - Amazon MQ | 216989113537 | Messaging/integration | T3 |
| 17 | Erajaya - ERM | 730335652953 | Utility/small | T3 |
| 18 | Erajaya - ESL | 356315792955 | Utility/small | T3 |
| 19 | Erajaya - EV | 574128098681 | Utility/small | T3 |
| 20 | Erajaya - Erfos | 058264280091 | Utility/small | T3 |
| 21 | Erajaya - POC - Quicksuite | 992382515728 | Utility/POC | T3 |
| 22 | Erajaya - POD | 730335345123 | Utility/small | T3 |
| 23 | Erajaya - Pentest | 194528794823 | Special/security | T3 |
| 24 | Erajaya - Rifos | 904869825152 | Utility/small | T3 |
| 25 | Erajaya - SAP Middleware | 273354665533 | Messaging/integration | T3 |
| 26 | Erajaya - SSO | 370924438980 | Utility/identity | T3 |
Note: 'Hypothesis Tier' column is initial estimate (T1? = deep dive candidate). Final tier determined after actual cost baseline. 'Small' accounts can be promoted if unexpectedly expensive (e.g., hidden data transfer).
Alur kerja assessment dengan kriteria selesai per tahap
Activities: Lock scope 26 accounts; activate read-only role; activate Compute Optimizer (by Erajaya); set 3-6 month period.
Output: Access & scope ready
Selesai jika: ICS Compute can view cost across all accounts.
Activities: Pull cost per Linked Account from payer; rank contribution; identify trends & anomalies.
Output: Cost Baseline & Tiering Report
Selesai jika: Final Tier 1/2/3 agreed (data-driven).
Activities: Breakdown service→usage type→resource; run optimization checklist; workshop with workload owners.
Output: Tier 1 findings validated
Selesai jika: Tier 1 findings validated by workload owners.
Activities: Tier 2: pattern-based quick wins + top-5 services. Tier 3: check active/dormant & consolidation candidates.
Output: Quick wins list
Selesai jika: Cross-tier 2/3 quick wins cataloged.
Activities: Analyze RI/SP at payer level; calculate shared-cost dependency; detect cross-account duplication.
Output: Commitment & Shared-Cost Analysis
Selesai jika: Commitment + shared cost map complete.
Activities: Build priority matrix (saving × effort × risk); 3-wave roadmap; executive presentation.
Output: Final Report & Roadmap
Selesai jika: Final report accepted by Erajaya.
Actions for specific conditions
Action: Run full happy path; accurate resource-level analysis down to usage type.
Result: Highest recommendation quality.
Action: Baseline & tiering proceed with Cost Explorer; deep dive uses Compute Optimizer + Trusted Advisor + CloudWatch; flag findings for refinement after data collection.
Result: Assessment proceeds; some recommendations labeled as estimates.
Action: Proceed with accessible accounts (prioritize T1 candidates); other accounts remain in baseline via payer data; create 'blocked accounts' list.
Result: Momentum maintained; no wait for full access.
Action: Rapid escalation to workload owner (no wait for phase completion); trace root cause (usage type/region/service); flag as high-priority finding.
Result: Large savings potential captured early.
Action: Promote account tier. Tiering is dynamic, follows data not hypothesis.
Result: Effort remains proportional to actual cost.
Action: Use estimate (traffic/tag proportion) with transparent assumptions; recommend tagging & cost allocation improvements.
Result: Dependency mapped though not 100% precise + structural recommendations.
Action: Make tagging governance a strategic recommendation; use proxy for analysis (resource name/account/region).
Result: Assessment proceeds + cost accountability foundation.
Deep dive tailored to account characteristics
| Workload Type | Example Accounts | Optimization Focus |
|---|---|---|
| Data / analytics platform | Databricks, Tableau, Barito (Kafka) | DBU/cluster sizing, auto-terminate idle cluster, storage tiering, MSK broker sizing. |
| Messaging / integration | Amazon MQ, SAP Middleware | Broker/instance sizing, idle queue, right-size. |
| Core business applications | 3PL, JDA, SCM, POS, POSOFF, ROP, SDP, TMS, Inbound, DRPPS | Right-sizing EC2/RDS, schedule non-prod (stop outside business hours), commitment. |
| AI / LLM | LLM Orchestrator | GPU/inference instance, on-demand vs commitment, idle endpoint. |
| Utility / small / special | Erfos, ERM, ESL, EV, Pentest, Rifos, SSO, POD, POC-Quicksuite, AMS | Check dormant, orphaned resource, consolidation/decommission. |
Indicative mapping; final tier & approach follows actual cost data.
| Risk | Impact | Mitigation |
|---|---|---|
| Read-only access approval delayed | Assessment delayed | Start from payer data + ready accounts (Scenario C). |
| Data not yet complete | Less granular recommendations | Activate early; use Cost Explorer + Compute Optimizer meanwhile (Scenario B). |
| Poor tagging | Difficult cost attribution | Use proxy + recommend tagging governance (Scenario G). |
| Idle misinterpreted | Incorrect recommendation (stopping critical resource) | Mandatory workshop validation before final recommendation. |
| Data spike masks trend | Biased baseline | Clean/annotate anomalies before baseline calculation. |
| Shared cost not separable | Imprecise dependency | Transparent estimate + allocation improvement recommendation (Scenario F). |
Cost contribution ranking per account; final tier assignment; identified trends & anomalies.
Resource-level breakdown per service & usage type; findings validated by workload owners; root cause analysis.
Pattern-based opportunities: gp2→gp3, unattached resources, snapshot retention, log retention, dormant checks.
RI/Savings Plans modeling at payer level; coverage optimization; commitment strategy.
Network/SSO/Shared resource cost allocation to Non-SAP workloads; duplication identified.
Priority matrix (saving × effort × risk); 3-wave roadmap; execution timeline.
High-level findings, total potential savings, risk assessment, strategic recommendations, next steps.
PT Erajaya Swasembada Tbk
AWS Cost Optimization Review - Non-SAP
ICS Compute
Infrastructure Managed Services
September 2026