service-capacity-modeling 0.3.100__py3-none-any.whl → 0.3.101__py3-none-any.whl

service_capacity_modeling/tools/capture_baseline_costs.py

@@ -0,0 +1,350 @@
+#!/usr/bin/env python3
+"""
+Capture current cost outputs for regression testing.
+
+This script runs capacity planning for various scenarios and captures
+the cost breakdowns to use as baselines for regression tests.
+
+Usage:
+    python -m service_capacity_modeling.tools.capture_baseline_costs
+"""
+
+import json
+from pathlib import Path
+from typing import Any
+
+from service_capacity_modeling.capacity_planner import planner
+from service_capacity_modeling.interface import (
+    AccessConsistency,
+    CapacityDesires,
+    certain_float,
+    certain_int,
+    Consistency,
+    DataShape,
+    GlobalConsistency,
+    Interval,
+    QueryPattern,
+)
+
+
+def capture_costs(
+    model_name: str,
+    region: str,
+    desires: CapacityDesires,
+    extra_args: dict[str, Any] | None = None,
+    scenario_name: str = "",
+) -> dict[str, Any]:
+    """Capture all cost breakdown for a planning scenario."""
+    try:
+        cap_plans = planner.plan_certain(
+            model_name=model_name,
+            region=region,
+            desires=desires,
+            num_results=1,
+            extra_model_arguments=extra_args or {},
+        )
+
+        if not cap_plans:
+            return {"error": "No capacity plans generated", "scenario": scenario_name}
+
+        cap_plan = cap_plans[0]
+        clusters = cap_plan.candidate_clusters
+
+        result = {
+            "scenario": scenario_name,
+            "model": model_name,
+            "region": region,
+            "service_tier": desires.service_tier,
+            "annual_costs": dict(
+                sorted((k, float(v)) for k, v in clusters.annual_costs.items())
+            ),
+            "total_annual_cost": float(clusters.total_annual_cost),
+            "cluster_count": len(clusters.zonal) + len(clusters.regional),
+            "service_count": len(clusters.services),
+        }
+
+        # Add instance info
+        if clusters.zonal:
+            result["instance_name"] = clusters.zonal[0].instance.name
+            result["instance_count"] = clusters.zonal[0].count
+            result["deployment"] = "zonal"
+        elif clusters.regional:
+            result["instance_name"] = clusters.regional[0].instance.name
+            result["instance_count"] = clusters.regional[0].count
+            result["deployment"] = "regional"
+
+        return result
+    except (ValueError, KeyError, AttributeError) as e:
+        return {"error": str(e), "scenario": scenario_name}
+
+
+# Define test scenarios for each service
+# Each scenario: (model_name, region, desires, extra_args, scenario_name)
+scenarios: list[tuple[str, str, CapacityDesires, dict[str, Any] | None, str]] = []
+
+# RDS scenarios
+rds_small = CapacityDesires(
+    service_tier=1,
+    query_pattern=QueryPattern(
+        estimated_read_per_second=certain_int(200),
+        estimated_write_per_second=certain_int(100),
+        estimated_mean_read_latency_ms=certain_float(10),
+        estimated_mean_write_latency_ms=certain_float(10),
+    ),
+    data_shape=DataShape(
+        estimated_state_size_gib=certain_int(50),
+    ),
+)
+
+rds_tier3 = CapacityDesires(
+    service_tier=3,
+    query_pattern=QueryPattern(
+        estimated_read_per_second=certain_int(200),
+        estimated_write_per_second=certain_int(100),
+        estimated_mean_read_latency_ms=certain_float(20),
+        estimated_mean_write_latency_ms=certain_float(20),
+    ),
+    data_shape=DataShape(
+        estimated_state_size_gib=certain_int(200),
+    ),
+)
+
+scenarios.extend(
+    [
+        ("org.netflix.rds", "us-east-1", rds_small, None, "rds_small_tier1"),
+        ("org.netflix.rds", "us-east-1", rds_tier3, None, "rds_tier3"),
+    ]
+)
+
+# Aurora scenarios
+aurora_small = CapacityDesires(
+    service_tier=1,
+    query_pattern=QueryPattern(
+        estimated_read_per_second=certain_int(100),
+        estimated_write_per_second=certain_int(100),
+        estimated_mean_read_latency_ms=certain_float(10),
+        estimated_mean_write_latency_ms=certain_float(10),
+    ),
+    data_shape=DataShape(
+        estimated_state_size_gib=certain_int(50),
+    ),
+)
+
+aurora_tier3 = CapacityDesires(
+    service_tier=3,
+    query_pattern=QueryPattern(
+        estimated_read_per_second=certain_int(200),
+        estimated_write_per_second=certain_int(100),
+        estimated_mean_read_latency_ms=certain_float(10),
+        estimated_mean_write_latency_ms=certain_float(10),
+    ),
+    data_shape=DataShape(
+        estimated_state_size_gib=certain_int(200),
+    ),
+)
+
+scenarios.extend(
+    [
+        ("org.netflix.aurora", "us-east-1", aurora_small, None, "aurora_small_tier1"),
+        ("org.netflix.aurora", "us-east-1", aurora_tier3, None, "aurora_tier3"),
+    ]
+)
+
+# Cassandra scenarios
+cassandra_small_high_qps = CapacityDesires(
+    service_tier=1,
+    query_pattern=QueryPattern(
+        estimated_read_per_second=certain_int(100_000),
+        estimated_write_per_second=certain_int(100_000),
+        estimated_mean_read_latency_ms=certain_float(0.5),
+        estimated_mean_write_latency_ms=certain_float(0.4),
+    ),
+    data_shape=DataShape(
+        estimated_state_size_gib=certain_int(10),
+    ),
+)
+
+cassandra_high_writes = CapacityDesires(
+    service_tier=1,
+    query_pattern=QueryPattern(
+        estimated_read_per_second=certain_int(10_000),
+        estimated_write_per_second=certain_int(500_000),
+    ),
+    data_shape=DataShape(
+        estimated_state_size_gib=certain_int(300),
+    ),
+)
+
+scenarios.extend(
+    [
+        (
+            "org.netflix.cassandra",
+            "us-east-1",
+            cassandra_small_high_qps,
+            {"require_local_disks": True},
+            "cassandra_small_high_qps_local",
+        ),
+        (
+            "org.netflix.cassandra",
+            "us-east-1",
+            cassandra_high_writes,
+            {"require_local_disks": False, "copies_per_region": 2},
+            "cassandra_high_writes_ebs",
+        ),
+    ]
+)
+
+# Kafka scenarios - Kafka uses throughput-based sizing via write_size
+# 100 MiB/s throughput with 2 consumers, 1 producer
+throughput = 100 * 1024 * 1024  # 100 MiB/s
+kafka_throughput = CapacityDesires(
+    service_tier=1,
+    query_pattern=QueryPattern(
+        estimated_read_per_second=Interval(low=1, mid=2, high=2, confidence=0.98),
+        estimated_write_per_second=Interval(low=1, mid=1, high=1, confidence=0.98),
+        estimated_mean_write_size_bytes=Interval(
+            low=throughput, mid=throughput, high=throughput * 2, confidence=0.98
+        ),
+    ),
+)
+
+scenarios.extend(
+    [
+        (
+            "org.netflix.kafka",
+            "us-east-1",
+            kafka_throughput,
+            {"require_local_disks": False},
+            "kafka_100mib_throughput",
+        ),
+    ]
+)
+
+# EVCache scenarios
+# Tiny EVCache - small cluster to show spread cost (< 10 instances = spread penalty)
+evcache_tiny = CapacityDesires(
+    service_tier=1,
+    query_pattern=QueryPattern(
+        estimated_read_per_second=certain_int(1_000),
+        estimated_write_per_second=certain_int(100),
+        estimated_mean_read_latency_ms=certain_float(1.0),
+    ),
+    data_shape=DataShape(
+        estimated_state_size_gib=certain_int(1),
+        estimated_state_item_count=Interval(
+            low=10_000, mid=100_000, high=200_000, confidence=0.98
+        ),
+    ),
+)
+
+evcache_small = CapacityDesires(
+    service_tier=1,
+    query_pattern=QueryPattern(
+        estimated_read_per_second=certain_int(100_000),
+        estimated_write_per_second=certain_int(10_000),
+        estimated_mean_read_latency_ms=certain_float(1.0),
+    ),
+    data_shape=DataShape(
+        estimated_state_size_gib=certain_int(10),
+        estimated_state_item_count=Interval(
+            low=1_000_000, mid=10_000_000, high=20_000_000, confidence=0.98
+        ),
+    ),
+)
+
+evcache_large = CapacityDesires(
+    service_tier=1,
+    query_pattern=QueryPattern(
+        estimated_read_per_second=certain_int(500_000),
+        estimated_write_per_second=certain_int(50_000),
+        estimated_mean_read_latency_ms=certain_float(1.0),
+    ),
+    data_shape=DataShape(
+        estimated_state_size_gib=certain_int(500),
+        estimated_state_item_count=Interval(
+            low=10_000_000, mid=100_000_000, high=200_000_000, confidence=0.98
+        ),
+    ),
+)
+
+scenarios.extend(
+    [
+        (
+            "org.netflix.evcache",
+            "us-east-1",
+            evcache_tiny,
+            {"cross_region_replication": "none"},
+            "evcache_tiny_with_spread",
+        ),
+        (
+            "org.netflix.evcache",
+            "us-east-1",
+            evcache_small,
+            {"cross_region_replication": "none"},
+            "evcache_small_no_replication",
+        ),
+        (
+            "org.netflix.evcache",
+            "us-east-1",
+            evcache_large,
+            {"cross_region_replication": "sets", "copies_per_region": 2},
+            "evcache_large_with_replication",
+        ),
+    ]
+)
+
+# Key-Value scenarios (composite: Cassandra + EVCache)
+# Uses evcache_large desires with eventual consistency to enable caching layer
+kv_with_cache = evcache_large.model_copy(deep=True)
+kv_with_cache.query_pattern.access_consistency = GlobalConsistency(
+    same_region=Consistency(target_consistency=AccessConsistency.eventual),
+    cross_region=Consistency(target_consistency=AccessConsistency.best_effort),
+)
+
+scenarios.extend(
+    [
+        (
+            "org.netflix.key-value",
+            "us-east-1",
+            kv_with_cache,
+            None,
+            "kv_with_cache",
+        ),
+    ]
+)
+
+# Export as dict for tests to import (single source of truth)
+SCENARIOS: dict[str, dict[str, Any]] = {
+    name: {
+        "model": model,
+        "region": region,
+        "desires": desires,
+        "extra_args": extra_args,
+    }
+    for model, region, desires, extra_args, name in scenarios
+}
+
+
+if __name__ == "__main__":
+    # Capture all scenarios
+    results = []
+    for model, region, desires, extra_args, scenario_name in scenarios:
+        print(f"Capturing: {scenario_name}...")
+        result = capture_costs(model, region, desires, extra_args, scenario_name)
+        results.append(result)
+
+        if "error" in result:
+            print(f"  ERROR: {result['error']}")
+        else:
+            print(f"  Total cost: ${result['total_annual_cost']:,.2f}")
+            print(f"  Cost breakdown: {list(result['annual_costs'].keys())}")
+
+    # Save results
+    output_file = Path(__file__).parent / "data" / "baseline_costs.json"
+    with open(output_file, "w", encoding="utf-8") as f:
+        json.dump(results, f, indent=2, sort_keys=True)
+        f.write("\n")  # Ensure trailing newline for pre-commit
+
+    print(f"\nResults saved to: {output_file}")
+    success_count = len([r for r in results if "error" not in r])
+    print(f"Total scenarios captured: {success_count}/{len(results)}")

{service_capacity_modeling-0.3.100.dist-info → service_capacity_modeling-0.3.101.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: service-capacity-modeling
-Version: 0.3.100
+Version: 0.3.101
 Summary: Contains utilities for modeling capacity for pluggable workloads
 Author: Joseph Lynch
 Author-email: josephl@netflix.com

{service_capacity_modeling-0.3.100.dist-info → service_capacity_modeling-0.3.101.dist-info}/RECORD

@@ -79,12 +79,14 @@ service_capacity_modeling/models/org/netflix/wal.py,sha256=QtRlqP_AIVpTg-XEINAfv
 service_capacity_modeling/models/org/netflix/zookeeper.py,sha256=T_CkmRqoEVqpERCFPU8xihyaxlNfUHDJXz7dMHM8GD0,7679
 service_capacity_modeling/tools/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 service_capacity_modeling/tools/auto_shape.py,sha256=K248-DayPrcZwLw1dYr47lpeQQwL0ylh1WAoVSdLNxw,23621
+service_capacity_modeling/tools/capture_baseline_costs.py,sha256=VXa5AEGpK4fGFhsDgKv-jKSkgLL8arAkuogv8Vm3Gq4,10729
 service_capacity_modeling/tools/fetch_pricing.py,sha256=fO84h77cqiiIHF4hZt490RwbZ6JqjB45UsnPpV2AXD4,6122
 service_capacity_modeling/tools/generate_missing.py,sha256=F7YqvMJAV4nZc20GNrlIsnQSF8_77sLgwYZqc5k4LDg,3099
 service_capacity_modeling/tools/instance_families.py,sha256=e5RuYkCLUITvsAazDH12B6KjX_PaBsv6Ne3mj0HK_sQ,9223
-service_capacity_modeling-0.3.100.dist-info/licenses/LICENSE,sha256=nl_Lt5v9VvJ-5lWJDT4ddKAG-VZ-2IaLmbzpgYDz2hU,11343
-service_capacity_modeling-0.3.100.dist-info/METADATA,sha256=LF_RV_duMAiGaOc7BX7jDNBjizo2BRtOX-cZtWLPYzw,10367
-service_capacity_modeling-0.3.100.dist-info/WHEEL,sha256=qELbo2s1Yzl39ZmrAibXA2jjPLUYfnVhUNTlyF1rq0Y,92
-service_capacity_modeling-0.3.100.dist-info/entry_points.txt,sha256=ZsjzpG5SomWpT1zCE19n1uSXKH2gTI_yc33sdl0vmJg,146
-service_capacity_modeling-0.3.100.dist-info/top_level.txt,sha256=H8XjTCLgR3enHq5t3bIbxt9SeUkUT8HT_SDv2dgIT_A,26
-service_capacity_modeling-0.3.100.dist-info/RECORD,,
+service_capacity_modeling/tools/data/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+service_capacity_modeling-0.3.101.dist-info/licenses/LICENSE,sha256=nl_Lt5v9VvJ-5lWJDT4ddKAG-VZ-2IaLmbzpgYDz2hU,11343
+service_capacity_modeling-0.3.101.dist-info/METADATA,sha256=6tuDUib_sRWz0Y9imRcO6t5KhF-qr4v1kXG8zbmewj0,10367
+service_capacity_modeling-0.3.101.dist-info/WHEEL,sha256=qELbo2s1Yzl39ZmrAibXA2jjPLUYfnVhUNTlyF1rq0Y,92
+service_capacity_modeling-0.3.101.dist-info/entry_points.txt,sha256=ZsjzpG5SomWpT1zCE19n1uSXKH2gTI_yc33sdl0vmJg,146
+service_capacity_modeling-0.3.101.dist-info/top_level.txt,sha256=H8XjTCLgR3enHq5t3bIbxt9SeUkUT8HT_SDv2dgIT_A,26
+service_capacity_modeling-0.3.101.dist-info/RECORD,,