forked from diegoecab/oci-deal-accelerator
69b0ccb4b8
Cross-referenced all 70+ KB files against official Oracle docs. Critical fixes: - ExaCS: storage was flash cache not usable disk (76.8→190.8 TB QR), added X11M, removed X10M - ADB-S: OCPU→ECPU migration, 26ai as default, cross-region RTO 15→10 min - Compute: E5 max 94→126, A2 80→78, added E6/A4/BM shapes, 11 GPU shapes - Storage: Higher Perf IOPS 35K→50K, UHP 450K→300K, File Storage 25K→50K - DBCS: E5.Flex max 94→64, storage 80→150 TB, added 26ai - Networking: FastConnect +400 Gbps, NSG max 120 rules Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
73 lines
2.6 KiB
YAML
73 lines
2.6 KiB
YAML
# CPU Conversion Ratios: vCPU → OCPU
|
|
# Used for right-sizing when migrating from other platforms
|
|
|
|
conversion_ratios:
|
|
description: |
|
|
OCI uses OCPUs (Oracle Compute Units). 1 OCPU = 1 physical core = 2 vCPUs.
|
|
When migrating from AWS/Azure/GCP, convert vCPUs to OCPUs using these ratios.
|
|
|
|
# Source → OCI OCPU conversion
|
|
aws:
|
|
general:
|
|
ratio: "2 AWS vCPUs = 1 OCI OCPU"
|
|
notes: "AWS vCPU = 1 hyperthread. OCI OCPU = 1 physical core (2 threads)."
|
|
by_processor:
|
|
intel_xeon_2nd_gen: # m5, c5, r5
|
|
vcpu_to_ocpu: 2.0
|
|
intel_xeon_3rd_gen: # m6i, c6i, r6i
|
|
vcpu_to_ocpu: 2.0
|
|
amd_epyc_2nd_gen: # m5a, c5a, r5a
|
|
vcpu_to_ocpu: 2.0
|
|
notes: "OCI E4 (AMD EPYC 3rd gen) is typically faster per core"
|
|
graviton2: # m6g, c6g, r6g
|
|
vcpu_to_ocpu: 1.0
|
|
notes: "Graviton2 vCPU = 1 physical core. Map 1:1 to Ampere A1."
|
|
graviton3: # m7g, c7g, r7g
|
|
vcpu_to_ocpu: 1.0
|
|
notes: "Map to Ampere A2 for comparable performance."
|
|
|
|
azure:
|
|
general:
|
|
ratio: "2 Azure vCores = 1 OCI OCPU (for Intel/AMD)"
|
|
by_processor:
|
|
intel_dv5:
|
|
vcpu_to_ocpu: 2.0
|
|
amd_dv5a:
|
|
vcpu_to_ocpu: 2.0
|
|
|
|
on_prem:
|
|
oracle_database:
|
|
description: "Use AWR Top Activity to determine actual core utilization"
|
|
approach: |
|
|
1. Get AWR "DB Time per second" for P75 interval
|
|
2. Divide by number of available cores
|
|
3. Result = core utilization ratio
|
|
4. Multiply by source cores to get required OCPUs
|
|
5. Apply 1.2x headroom factor
|
|
fallback: "If no AWR data: 1 on-prem core = 1 OCPU (conservative)"
|
|
vmware:
|
|
ratio: "2 VMware vCPUs = 1 OCI OCPU"
|
|
notes: "VMware vCPU is a hyperthread equivalent"
|
|
|
|
adb_s_sizing:
|
|
description: |
|
|
For ADB-S, size base OCPUs for P75 sustained load:
|
|
- Auto-scaling can go up to 3x base OCPUs
|
|
- Auto-scaling activation takes 2-3 minutes
|
|
- If P95 spikes are short (< 2 min), they may hit before scale-out
|
|
- Size base for P75 to ensure good baseline performance
|
|
|
|
formula: |
|
|
base_ocpus = ceil(p75_cpu_usage / cores_per_ocpu * headroom_factor)
|
|
headroom_factor = 1.2 (default)
|
|
auto_scale_max = base_ocpus * 3
|
|
|
|
memory_sizing:
|
|
oci_shapes:
|
|
standard_e4: "16 GB per OCPU (VM.Standard.E4.Flex)"
|
|
standard_e5: "up to 64 GB per OCPU (VM.Standard.E5.Flex, default 16)"
|
|
standard_a1: "6 GB per OCPU (VM.Standard.A1.Flex / Ampere)"
|
|
standard_a2: "up to 64 GB per OCPU (VM.Standard.A2.Flex / Ampere)"
|
|
optimized3: "14 GB per OCPU (VM.Optimized3.Flex)"
|
|
dense_io_e4: "32 GB per OCPU + local NVMe (VM.DenseIO.E4.Flex)"
|