introduced failure manager

Author: networmix

ngraph/failure_manager.py

@@ -0,0 +1,192 @@
+from __future__ import annotations
+
+import copy
+import statistics
+from collections import defaultdict
+from concurrent.futures import ThreadPoolExecutor, as_completed
+from typing import List, Dict, Optional, Tuple, Any
+
+from ngraph.network import Network
+from ngraph.traffic_demand import TrafficDemand
+from ngraph.traffic_manager import TrafficManager, TrafficResult
+from ngraph.failure_policy import FailurePolicy
+
+
+class FailureManager:
+    """
+    Applies FailurePolicy to a Network, runs traffic placement, and (optionally)
+    repeats multiple times for Monte Carlo experiments.
+
+    Attributes:
+        network (Network): The underlying network to mutate (enable/disable nodes/links).
+        traffic_demands (List[TrafficDemand]): List of demands to place after failures.
+        failure_policy (Optional[FailurePolicy]): The policy describing what fails.
+        default_flow_policy_config: The default flow policy for any demands lacking one.
+    """
+
+    def __init__(
+        self,
+        network: Network,
+        traffic_demands: List[TrafficDemand],
+        failure_policy: Optional[FailurePolicy] = None,
+        default_flow_policy_config=None,
+    ) -> None:
+        """
+        Initialize a FailureManager.
+
+        Args:
+            network: The Network to be modified by failures.
+            traffic_demands: Demands to place on the network after applying failures.
+            failure_policy: A FailurePolicy specifying the rules of what fails.
+            default_flow_policy_config: Default FlowPolicyConfig if demands do not specify one.
+        """
+        self.network = network
+        self.traffic_demands = traffic_demands
+        self.failure_policy = failure_policy
+        self.default_flow_policy_config = default_flow_policy_config
+
+    def apply_failures(self) -> None:
+        """
+        Apply the current failure_policy to self.network (in-place).
+
+        If failure_policy is None, this method does nothing.
+        """
+        if not self.failure_policy:
+            return
+
+        # Collect node/links as dicts {id: attrs}, matching FailurePolicy expectations
+        node_map = {n_name: n.attrs for n_name, n in self.network.nodes.items()}
+        link_map = {l_id: l.attrs for l_id, l in self.network.links.items()}
+
+        failed_ids = self.failure_policy.apply_failures(node_map, link_map)
+
+        # Disable the failed entities
+        for f_id in failed_ids:
+            if f_id in self.network.nodes:
+                self.network.disable_node(f_id)
+            elif f_id in self.network.links:
+                self.network.disable_link(f_id)
+
+    def run_single_failure_scenario(self) -> List[TrafficResult]:
+        """
+        Applies failures to the network, places the demands, and returns per-demand results.
+
+        Returns:
+            List[TrafficResult]: A list of traffic result objects under the applied failures.
+        """
+        # Ensure we start with a fully enabled network (in case of reuse)
+        self.network.enable_all()
+
+        # Apply the current failure policy
+        self.apply_failures()
+
+        # Build TrafficManager and place demands
+        tmgr = TrafficManager(
+            network=self.network,
+            traffic_demands=copy.deepcopy(self.traffic_demands),
+            default_flow_policy_config=self.default_flow_policy_config,
+        )
+        tmgr.build_graph()
+        tmgr.expand_demands()
+        tmgr.place_all_demands()
+
+        # Return detailed traffic results
+        return tmgr.get_traffic_results(detailed=True)
+
+    def run_monte_carlo_failures(
+        self,
+        iterations: int,
+        parallelism: int = 1,
+    ) -> Dict[str, Any]:
+        """
+        Repeatedly applies (randomized) failures to the network and accumulates
+        per-run traffic data. Returns both overall volume statistics and a
+        breakdown of results for each (src, dst, priority).
+
+        Args:
+            iterations (int): Number of times to run the failure scenario.
+            parallelism (int): Max number of worker threads to use (for parallel runs).
+
+        Returns:
+            Dict[str, Any]: A dictionary containing:
+                {
+                    "overall_stats": {
+                        "mean": <float>,
+                        "stdev": <float>,
+                        "min": <float>,
+                        "max": <float>
+                    },
+                    "by_src_dst": {
+                        (src, dst, priority): [
+                            {
+                                "iteration": <int>,
+                                "total_volume": <float>,
+                                "placed_volume": <float>,
+                                "unplaced_volume": <float>
+                            },
+                            ...
+                        ],
+                        ...
+                    }
+                }
+        """
+        # scenario_list will hold the list of traffic-results (List[TrafficResult]) per iteration
+        scenario_list: List[List[TrafficResult]] = []
+
+        # Run in parallel or synchronously
+        if parallelism > 1:
+            with ThreadPoolExecutor(max_workers=parallelism) as executor:
+                futures = [
+                    executor.submit(self.run_single_failure_scenario)
+                    for _ in range(iterations)
+                ]
+                for f in as_completed(futures):
+                    scenario_list.append(f.result())
+        else:
+            for _ in range(iterations):
+                scenario_list.append(self.run_single_failure_scenario())
+
+        # If no scenarios were run, return zeroed stats
+        if not scenario_list:
+            return {
+                "overall_stats": {"mean": 0.0, "stdev": 0.0, "min": 0.0, "max": 0.0},
+                "by_src_dst": {},
+            }
+
+        # Accumulate total placed volumes for each iteration (for top-level summary)
+        placed_totals: List[float] = []
+
+        # Dictionary mapping (src, dst, priority) -> list of run-by-run results
+        by_src_dst: Dict[Tuple[str, str, int], List[Dict[str, float]]] = defaultdict(
+            list
+        )
+
+        for i, traffic_results in enumerate(scenario_list):
+            # Compute total placed volume for this iteration
+            scenario_placed_total = sum(r.placed_volume for r in traffic_results)
+            placed_totals.append(scenario_placed_total)
+
+            # Accumulate detailed data for each (src, dst, priority)
+            for r in traffic_results:
+                key = (r.src, r.dst, r.priority)
+                by_src_dst[key].append(
+                    {
+                        "iteration": i,
+                        "total_volume": r.total_volume,
+                        "placed_volume": r.placed_volume,
+                        "unplaced_volume": r.unplaced_volume,
+                    }
+                )
+
+        # Compute overall statistics on the total placed volumes
+        overall_stats = {
+            "mean": statistics.mean(placed_totals),
+            "stdev": statistics.pstdev(placed_totals),
+            "min": min(placed_totals),
+            "max": max(placed_totals),
+        }
+
+        return {
+            "overall_stats": overall_stats,
+            "by_src_dst": dict(by_src_dst),
+        }