Skip to content

Integration Architecture Specifications

Edge AI + Network + Observability Platforms

This document provides the complete integration architecture showing how Cisco Unified Edge (UCS XE9305 + XE130c M8) integrates with: 1. Camera infrastructure (120 cameras per site) 2. Cisco network infrastructure (Catalyst 9300/9500, ISE, FTD) 3. AI observability platforms (Splunk MLTK, ThousandEyes, AppDynamics) 4. Security response systems (XDR, ServiceNow, Webex)


ARCHITECTURE OVERVIEW

┌─────────────────────────────────────────────────────────────────────────┐
│                 ABHAVTECH EDGE AI INTEGRATION ARCHITECTURE              │
│                    Mumbai & Chennai Hub Sites (Identical)               │
└─────────────────────────────────────────────────────────────────────────┘

PHYSICAL LAYER: Camera Infrastructure
┌──────────────────────────────────────────────────────────────────────┐
│ 120 Cameras per Site (Mumbai/Chennai)                                │
│ ├─ 60× Indoor Fixed (Axis P3715-PLVE): Hallways, conference rooms   │
│ ├─ 40× Outdoor PTZ (Axis Q6215-LE): Perimeter, 360° coverage        │
│ ├─ 20× 4K LPR (Axis P1455-LE): Vehicle entry/exit, parking          │
│ └─ 10× Thermal (FLIR A310f): Server room, electrical room fire      │
│                                                                       │
│ Configuration:                                                        │
│ ├─ VLAN: VN_IOT (150), IPs: 10.150.2.0/24 - 10.150.9.0/24          │
│ ├─ SGT: SGT-70 (Cameras) - assigned by ISE via 802.1X/MAC          │
│ ├─ Protocol: RTSP over TCP (port 554)                               │
│ ├─ Codec: H.265 (HEVC) adaptive bitrate 4-10 Mbps                  │
│ ├─ Power: PoE+ 802.3at (25-30W per camera)                         │
│ └─ Total Bandwidth: 120 cameras × 8 Mbps avg = 960 Mbps            │
└──────────────────────────────────────────────────────────────────────┘
                 1 Gbps Ethernet (PoE+ injection)
ACCESS LAYER: Catalyst 9300 Switches
┌──────────────────────────────────────────────────────────────────────┐
│ 6× Cisco Catalyst 9300-48U per Site                                  │
│ ├─ 48 ports × 1G RJ45 (PoE+, 1,100W budget per switch)             │
│ ├─ Camera distribution: 20 cameras per switch                       │
│ ├─ PoE consumption: 20 cameras × 25W = 500W (45% utilization)      │
│ ├─ Bandwidth per switch: 160 Mbps camera traffic                   │
│ ├─ SGACL enforcement: SGT-70→SGT-95 PERMIT, SGT-70→SGT-10 DENY    │
│ └─ Uplinks: 4× 10G SFP+ fiber (LAG) to Catalyst 9500               │
└──────────────────────────────────────────────────────────────────────┘
                 4× 10 Gbps Fiber (40 Gbps aggregate per switch)
DISTRIBUTION LAYER: Catalyst 9500 Switch
┌──────────────────────────────────────────────────────────────────────┐
│ 1× Cisco Catalyst 9500-40X per Site (IDF Room, Floor 3)             │
│ ├─ 40× 10G SFP+ ports (fiber)                                       │
│ ├─ Inter-VLAN routing: VN_IOT (150) → Edge AI                      │
│ ├─ Camera uplinks: 6 switches × 40 Gbps = 240 Gbps aggregate       │
│ ├─ Camera bandwidth: 960 Mbps (0.4% of 240 Gbps capacity)          │
│ ├─ Edge AI uplinks: 2× 10G LAG per node = 20 Gbps per node        │
│ │   ├─ Primary node: Ports 1-2 (20 Gbps)                           │
│ │   └─ Standby node: Ports 3-4 (20 Gbps)                           │
│ ├─ WAN uplinks: 2× 10G to core router (observability APIs)        │
│ └─ Routing: OSPF Area 0 (internal), BGP AS 65000 (WAN to NJ)      │
└──────────────────────────────────────────────────────────────────────┘
                 2× 10 Gbps Fiber (Link Aggregation = 20 Gbps)
EDGE AI LAYER: Cisco Unified Edge
┌──────────────────────────────────────────────────────────────────────┐
│ Cisco UCS XE9305 Chassis (3 RU, 18" short-depth)                    │
│ Location: IDF Room, Floor 3 (co-located with Catalyst switches)     │
│ ├─ Form Factor: 3 RU rack mount, 40 lbs weight                     │
│ ├─ Compute Slots: 5 slots (2 used, 3 reserved for Phase 5)         │
│ ├─ Power: Dual 1,000W PSU (N+1), 700W actual consumption           │
│ ├─ Fabric: Embedded 25G switches (2× management controllers)       │
│ ├─ Cooling: 5× hot-swap fans (60dB, IDF-optimized acoustics)      │
│ └─ Management: Cisco Intersight Infrastructure Service (SaaS)       │
│                                                                       │
│ ┌─────────────────────────────────────────────────────────────────┐ │
│ │ SLOT 1 (Primary): UCS XE130c M8 Compute Node                    │ │
│ │ ───────────────────────────────────────────────────────────────  │ │
│ │ Hostname: edge-ai-mumbai-01                                      │ │
│ │ Management IP: 10.150.1.10                                       │ │
│ │ VRRP VIP: 10.150.1.1 (active, shared with standby)             │ │
│ │ SGT: SGT-95 (Edge AI Servers)                                   │ │
│ │                                                                   │ │
│ │ CPU: Intel Xeon 6 SoC (32 cores @ 2.6 GHz base, 4.0 GHz turbo) │ │
│ │ RAM: 128GB DDR5-4800 (4 channels × 32GB DIMMs)                  │ │
│ │ GPU: NVIDIA L4 24GB GDDR6 (PCIe Gen5 slot, 72W TDP)            │ │
│ │   └─ Performance: 120 TOPS INT8 inference                       │ │
│ │   └─ Utilization: 70-80% target (120 cameras @ 30 FPS)         │ │
│ │ Storage:                                                          │ │
│ │   ├─ Boot: Dual M.2 SATA 512GB (RAID 1)                        │ │
│ │   └─ Data: 2× E3.S NVMe 1TB (7-day event buffer)               │ │
│ │ Network:                                                          │ │
│ │   ├─ Midplane: 2× 25G (chassis fabric, management)             │ │
│ │   └─ Front-panel: 2× 10G SFP+ (LAG to Catalyst 9500)           │ │
│ │ Power: 350W avg (CPU 185W + GPU 72W + other 93W)               │ │
│ │                                                                   │ │
│ │ AI Workload:                                                      │ │
│ │ ├─ YOLO v8n: Person detection (20ms inference)                  │ │
│ │ ├─ DeepSORT: Object tracking (loitering, 10ms overhead)        │ │
│ │ ├─ Custom CNN: PPE detection (hard hat, safety vest, 25ms)     │ │
│ │ └─ LPR Pipeline: YOLO + CNN + Tesseract (127ms total)          │ │
│ │                                                                   │ │
│ │ Inbound: 960 Mbps (120 camera RTSP streams)                     │ │
│ │ Outbound: 50 Mbps (observability API calls)                     │ │
│ │ Latency: 4ms camera → edge AI (same IDF room)                  │ │
│ └─────────────────────────────────────────────────────────────────┘ │
│                                                                       │
│ ┌─────────────────────────────────────────────────────────────────┐ │
│ │ SLOT 2 (Standby): UCS XE130c M8 Compute Node                    │ │
│ │ ───────────────────────────────────────────────────────────────  │ │
│ │ Hostname: edge-ai-mumbai-02                                      │ │
│ │ Management IP: 10.150.1.11                                       │ │
│ │ VRRP VIP: 10.150.1.1 (standby, takes over on failover)         │ │
│ │ Hardware: Identical to Slot 1                                    │ │
│ │ Role: Hot standby (5-sec heartbeat, RTO <30 sec)               │ │
│ │ GPU: Warm standby (models pre-loaded, ready for instant use)   │ │
│ └─────────────────────────────────────────────────────────────────┘ │
│                                                                       │
│ Slots 3-5: Reserved for Phase 5 (6 branch sites, 2 branches/node)  │
└──────────────────────────────────────────────────────────────────────┘
              Outbound: 50 Mbps (observability/security APIs)
OBSERVABILITY & SECURITY INTEGRATION
┌──────────────────────────────────────────────────────────────────────┐
│ Mumbai Datacenter (LAN, <50ms latency)                               │
│ ├─ ISE pxGrid (10.30.0.1:8910)                                      │
│ │   └─ Badge swipe correlation, ~5 Mbps, WebSocket                 │
│ ├─ FTD Firewall (ftd-mumbai.abhavtech.com)                         │
│ │   └─ Automated network blocking, ~2 Mbps, HTTPS API              │
│ └─ BMS Honeywell (bms.abhavtech.com)                               │
│     └─ HVAC/lighting control, ~1 Mbps, OAuth 2.0 API               │
│                                                                       │
│ NJ Datacenter (WAN, ~100ms latency, 10 Gbps MPLS)                   │
│ └─ Splunk MLTK (10.182.1.50:8089)                                   │
│     └─ Historical pattern validation, ~15 Mbps, HTTPS API           │
│                                                                       │
│ SaaS Cloud (Internet, ~80ms latency, 1 Gbps circuit)                │
│ ├─ ThousandEyes (api.thousandeyes.com)                             │
│ │   └─ Network path health, ~10 Mbps, OAuth 2.0                    │
│ ├─ AppDynamics (abhavtech.saas.appdynamics.com)                    │
│ │   └─ Application health, ~10 Mbps, API token auth                │
│ ├─ XDR SecureX (securex.cisco.com)                                  │
│ │   └─ Security incident correlation, ~2 Mbps, Bearer token        │
│ ├─ ServiceNow (abhavtech.service-now.com)                          │
│ │   └─ Incident ticketing, ~3 Mbps, Basic auth                     │
│ └─ Webex Teams (webexapis.com)                                      │
│     └─ Supervisor mobile alerts, ~2 Mbps, Bearer token             │
└──────────────────────────────────────────────────────────────────────┘

DETAILED DATA FLOW: PERIMETER INTRUSION DETECTION

Timeline: 0ms → 500ms (Detection → Supervisor Alert)

TIME: 14:32:05.000 - Frame Acquisition
┌──────────────────────────────────────────────────────────────┐
│ Camera 47 (Outdoor PTZ, Loading Dock Perimeter)              │
│ IP: 10.150.5.47, SGT: SGT-70                                 │
│ RTSP URL: rtsp://10.150.5.47:554/axis-media/media.amp       │
│ Frame: 1920×1080 @ 30 FPS, H.265, 8.2 Mbps current          │
│ Timestamp: 2025-01-15T14:32:05.000Z                         │
└──────────────────────────────────────────────────────────────┘
         ↓ 1 Gbps Ethernet, PoE+ 30W, <1ms latency
┌──────────────────────────────────────────────────────────────┐
│ Catalyst 9300-48U Access Switch #5                           │
│ Port 20, VLAN 150, SGT-70 assigned via ISE                  │
│ SGACL: Permit SGT-70 → SGT-95 (camera to edge AI)           │
│ QoS: AF31 (DSCP 26) for camera RTSP traffic                 │
└──────────────────────────────────────────────────────────────┘
         ↓ 10 Gbps Fiber (LAG, 4 ports), <1ms latency
┌──────────────────────────────────────────────────────────────┐
│ Catalyst 9500-40X Distribution Switch                        │
│ Inter-VLAN routing: 10.150.5.47 → 10.150.1.1 (VRRP VIP)    │
│ Route: OSPF Area 0, next-hop direct (same IDF room)         │
└──────────────────────────────────────────────────────────────┘
         ↓ 2× 10 Gbps Fiber (LAG), <2ms latency
┌──────────────────────────────────────────────────────────────┐
│ UCS XE130c M8 Primary Node (edge-ai-mumbai-01)              │
│ VRRP VIP: 10.150.1.1 (active), SGT: SGT-95                  │
│ Frame received at network interface (2× 10G LAG)             │
│ Total network latency: Camera → Edge AI = 4ms               │
└──────────────────────────────────────────────────────────────┘

TIME: 14:32:05.004 - Edge AI Processing Begins

┌──────────────────────────────────────────────────────────────┐
│ Container: AI Inference Service (K8s pod)                     │
│ RTSP Client: Pulls frame from camera buffer (in-memory)      │
│ Frame decoded: H.265 → raw RGB (1920×1080, 6.2 MB)          │
└──────────────────────────────────────────────────────────────┘
         ↓ 5ms - CPU pre-processing
┌──────────────────────────────────────────────────────────────┐
│ CPU Pre-Processing (Intel Xeon 6, 32 cores)                  │
│ ├─ Resize: 1920×1080 → 640×640 (letterbox padding)          │
│ ├─ Normalize: [0-255] → [0-1] float32                       │
│ ├─ Colorspace: BGR → RGB                                     │
│ └─ Transpose: HWC → CHW format (PyTorch input)              │
│ Output: 640×640×3 tensor ready for GPU                       │
└──────────────────────────────────────────────────────────────┘
         ↓ Copy tensor to GPU memory (PCIe Gen5, <1ms)
┌──────────────────────────────────────────────────────────────┐
│ GPU Inference (NVIDIA L4 24GB, PCIe Gen5 slot)              │
│ Model: YOLO v8n (INT8 quantized, 6.2 MB, TensorRT)          │
│ Input: 640×640×3 tensor                                      │
│ GPU Execution: 20ms @ 75% utilization                        │
│ Output: Bounding boxes, classes, confidences                 │
│ Result: 1 person detected at [850,320,970,600], conf=0.96   │
└──────────────────────────────────────────────────────────────┘
         ↓ 3ms - CPU post-processing
┌──────────────────────────────────────────────────────────────┐
│ CPU Post-Processing                                           │
│ ├─ Non-Max Suppression: Filter overlapping boxes             │
│ ├─ Zone validation: Bbox center in restricted zone polygon   │
│ └─ Duplicate check: Query SQLite (no duplicate last 60 sec)  │
│ Decision: NEW detection in RZ-LOADING-DOCK zone              │
└──────────────────────────────────────────────────────────────┘

TIME: 14:32:05.030 - Multi-Source Validation Launch

┌──────────────────────────────────────────────────────────────┐
│ Validation Orchestrator (Python async, 4 parallel API calls) │
└──────────────────────────────────────────────────────────────┘
         ↓ Parallel execution (max latency: 100ms)
┌──────────────────────────────────────────────────────────────┐
│ API Call 1: ISE pxGrid (Mumbai Datacenter)                   │
│ ───────────────────────────────────────────────────────────   │
│ Endpoint: https://10.30.0.1:8910/pxgrid/ise/session/query   │
│ Network: Mumbai LAN (same building)                          │
│ Request: Query badge swipes at Loading Dock (last 5 min)    │
│ Response Time: 50ms                                          │
│ Result: 0 badge swipes (UNAUTHORIZED entry)                 │
└──────────────────────────────────────────────────────────────┘

┌──────────────────────────────────────────────────────────────┐
│ API Call 2: Splunk MLTK (NJ Datacenter)                      │
│ ───────────────────────────────────────────────────────────   │
│ Endpoint: https://10.182.1.50:8089/services/search/jobs     │
│ Network: MPLS WAN Mumbai → NJ (10 Gbps, 200ms RTT)          │
│ Request: Historical occupancy pattern (Tuesday 14:30)        │
│ Response Time: 100ms                                          │
│ Result: Expected 0.2 people, Actual 1 (ANOMALOUS)           │
└──────────────────────────────────────────────────────────────┘

┌──────────────────────────────────────────────────────────────┐
│ API Call 3: ThousandEyes (SaaS Cloud)                        │
│ ───────────────────────────────────────────────────────────   │
│ Endpoint: https://api.thousandeyes.com/v6/tests/12345/...   │
│ Network: Internet (1 Gbps circuit)                           │
│ Request: Camera 47 → Edge AI network path health            │
│ Response Time: 80ms                                           │
│ Result: 0% loss, 12ms latency (NETWORK HEALTHY)             │
└──────────────────────────────────────────────────────────────┘

┌──────────────────────────────────────────────────────────────┐
│ API Call 4: AppDynamics (SaaS Cloud)                         │
│ ───────────────────────────────────────────────────────────   │
│ Endpoint: https://abhavtech.saas.appdynamics.com/...        │
│ Network: Internet (1 Gbps circuit)                           │
│ Request: RTSP service health (Camera 47 stream)              │
│ Response Time: 90ms                                           │
│ Result: 0% error rate (APPLICATION HEALTHY)                 │
└──────────────────────────────────────────────────────────────┘

TIME: 14:32:05.135 - Validation Complete (All 4 APIs responded)

┌──────────────────────────────────────────────────────────────┐
│ Decision Logic (Python decision engine)                      │
│ ───────────────────────────────────────────────────────────   │
│ Criteria:                                                     │
│ ✓ AI confidence: 96% (≥90% threshold)                       │
│ ✓ ISE: Unauthorized (0 badge swipes)                        │
│ ✓ Splunk: Anomalous (occupancy 5× expected)                 │
│ ✓ ThousandEyes: Network healthy (0% loss)                   │
│ ✓ AppDynamics: Application healthy (0% errors)              │
│                                                               │
│ Decision: HIGH CONFIDENCE (all 5 criteria passed)           │
│ Action: AUTOMATED response authorized                        │
└──────────────────────────────────────────────────────────────┘

TIME: 14:32:05.155 - Automated Actions (4 parallel API calls)

┌──────────────────────────────────────────────────────────────┐
│ Action 1: FTD Firewall Network Block (Mumbai LAN)            │
│ ───────────────────────────────────────────────────────────   │
│ Endpoint: https://ftd-mumbai.abhavtech.com/api/...          │
│ Request: Create block rule VLAN 150 → Corporate (30 min)    │
│ Response Time: 50ms                                          │
│ Result: Rule ID 005056BB...400 created, expires 15:02       │
└──────────────────────────────────────────────────────────────┘

┌──────────────────────────────────────────────────────────────┐
│ Action 2: XDR SecureX Incident (SaaS Cloud)                  │
│ ───────────────────────────────────────────────────────────   │
│ Endpoint: https://securex.cisco.com/iroh/...                │
│ Request: Create security incident (P2-High priority)         │
│ Response Time: 40ms                                          │
│ Result: Incident ID incident-2025-0147 created              │
└──────────────────────────────────────────────────────────────┘

┌──────────────────────────────────────────────────────────────┐
│ Action 3: ServiceNow Ticket (SaaS Cloud)                     │
│ ───────────────────────────────────────────────────────────   │
│ Endpoint: https://abhavtech.service-now.com/api/...         │
│ Request: Create incident with video snapshot attachment      │
│ Response Time: 60ms                                          │
│ Result: INC0012345 created, assigned to supervisor          │
└──────────────────────────────────────────────────────────────┘

┌──────────────────────────────────────────────────────────────┐
│ Action 4: Webex Teams Alert (SaaS Cloud)                     │
│ ───────────────────────────────────────────────────────────   │
│ Endpoint: https://webexapis.com/v1/messages                 │
│ Request: Send mobile push notification to supervisor         │
│ Response Time: 80ms                                          │
│ Result: Message sent, mobile push delivered in ~2-3 sec     │
└──────────────────────────────────────────────────────────────┘

TIME: 14:32:05.500 - Supervisor Receives Mobile Notification

TOTAL LATENCY: 500ms (Detection → Mobile Alert)
  ├─ Frame acquisition: 0ms
  ├─ Network (camera → edge AI): 4ms
  ├─ Pre-processing: 5ms
  ├─ GPU inference: 20ms
  ├─ Post-processing + duplicate check: 5ms
  ├─ Multi-source validation (parallel): 100ms
  ├─ Decision logic: 20ms
  ├─ Automated actions (parallel): 80ms
  └─ Network propagation (FTD + Webex push): 266ms

INTEGRATION ARCHITECTURE: API SPECIFICATIONS

ISE pxGrid Integration (Badge Event Correlation)

Purpose: Correlate person detections with badge swipe events to distinguish authorized vs. unauthorized entry.

Network Path:

Edge AI (10.150.1.10) → Catalyst 9500 → Core Router → ISE (10.30.0.1)
Latency: ~50ms (Mumbai LAN, same datacenter)
Bandwidth: ~5 Mbps (event stream, WebSocket persistent connection)

API Specification:

POST https://10.30.0.1:8910/pxgrid/ise/session/query
Authorization: Basic <base64_credentials>
Content-Type: application/json
X-Request-ID: perimeter-intrusion-20250115-143205

{
  "location": "Loading Dock",
  "startTimestamp": "2025-01-15T14:27:05.000Z",
  "endTimestamp": "2025-01-15T14:32:05.000Z",
  "sgt": ["SGT-71"]
}

Response:

{
  "requestId": "perimeter-intrusion-20250115-143205",
  "timestamp": "2025-01-15T14:32:05.085Z",
  "sessions": [],
  "totalCount": 0
}

Validation Logic: - totalCount == 0 → UNAUTHORIZED (no badge swipes detected) - totalCount >= 1 → AUTHORIZED (badge swipe detected, likely employee)


Splunk MLTK Integration (Historical Pattern Validation)

Purpose: Compare current occupancy vs. historical patterns to detect anomalies.

Network Path:

Edge AI (10.150.1.10) → Catalyst 9500 → Core Router → MPLS WAN → NJ Datacenter → Splunk (10.182.1.50)
Latency: ~100ms (200ms RTT Mumbai ↔ NJ, 10 Gbps MPLS circuit)
Bandwidth: ~15 Mbps (search queries + HEC event ingestion)

API Specification:

POST https://10.182.1.50:8089/services/search/jobs
Authorization: Bearer <splunk_token>
Content-Type: application/x-www-form-urlencoded

search=index=bms location="Loading Dock" earliest=-30d latest=now 
| eval hour=strftime(_time, "%H"), day=strftime(_time, "%A")
| where hour="14" AND day="Tuesday"
| stats avg(occupancy_count) as expected_occupancy
| eval current_occupancy=1
| eval anomaly=if(current_occupancy > expected_occupancy * 1.5 OR current_occupancy < expected_occupancy * 0.5, 1, 0)
&output_mode=json

Response:

{
  "sid": "1705317125.12345",
  "results": [
    {
      "expected_occupancy": 0.2,
      "current_occupancy": 1,
      "anomaly": 1
    }
  ]
}

Validation Logic: - anomaly == 1 → ANOMALOUS (occupancy significantly different from historical pattern) - anomaly == 0 → NORMAL (occupancy within expected range)


ThousandEyes Integration (Network Path Health)

Purpose: Validate camera → edge AI network path to rule out false positives from packet loss.

Network Path:

Edge AI (10.150.1.10) → Catalyst 9500 → Core Router → Internet (1 Gbps) → ThousandEyes Cloud
Latency: ~80ms (Mumbai → ThousandEyes Singapore region)
Bandwidth: ~10 Mbps (API queries, periodic polling)

API Specification:

GET https://api.thousandeyes.com/v6/tests/12345/net/path-vis
Authorization: Bearer <thousandeyes_token>
Content-Type: application/json

Parameters:
  testId: 12345
  window: 2m
  aid: 98765

Response:

{
  "test": {
    "testId": 12345,
    "testName": "Camera-47 → Edge-AI-Mumbai-01 Network Path"
  },
  "net": {
    "metrics": [
      {
        "date": "2025-01-15T14:32:00",
        "avgLatency": 12,
        "minLatency": 10,
        "maxLatency": 15,
        "loss": 0.0,
        "jitter": 2
      }
    ]
  }
}

Validation Logic: - loss < 1% AND latency < 100ms → NETWORK HEALTHY - loss >= 1% OR latency >= 100ms → NETWORK DEGRADED (may cause false positive)


AppDynamics Integration (Application Health)

Purpose: Validate RTSP streaming service health to rule out false positives from corrupt video.

Network Path:

Edge AI (10.150.1.10) → Catalyst 9500 → Core Router → Internet (1 Gbps) → AppDynamics Cloud
Latency: ~90ms (Mumbai → AppDynamics US East region)
Bandwidth: ~10 Mbps (metrics queries, periodic polling)

API Specification:

GET https://abhavtech.saas.appdynamics.com/controller/rest/applications/10/metric-data
Authorization: Bearer <appdynamics_token>
Content-Type: application/json

Parameters:
  application: 10
  metric-path: Business Transaction Performance|...|Camera-47-Stream|Errors per Minute
  time-range-type: BEFORE_NOW
  duration-in-mins: 5
  output: JSON

Response:

{
  "metric-data": [
    {
      "metricName": "BTM|BTs|BT:12345|...|Errors per Minute",
      "metricValues": [
        {"startTimeInMillis": 1705317120000, "value": 0, "count": 1}
      ]
    }
  ]
}

Validation Logic: - avg_error_rate < 5% → APPLICATION HEALTHY - avg_error_rate >= 5% → APPLICATION UNHEALTHY (may cause false positive)


SECURITY INTEGRATION: FTD + XDR + ServiceNow + Webex

FTD Firewall Automated Blocking

Purpose: Isolate compromised VLAN on high-confidence security events.

Network Path:

Edge AI (10.150.1.10) → Catalyst 9500 → Core Router → FTD (ftd-mumbai.abhavtech.com)
Latency: ~50ms (Mumbai LAN, same datacenter)
Bandwidth: ~2 Mbps (rule creation commands, low volume)

API Request:

POST https://ftd-mumbai.abhavtech.com/api/fmc_config/v1/domain/default/policy/accesspolicies/.../accessrules
Authorization: Bearer <ftd_token>
Content-Type: application/json

{
  "name": "BLOCK-LoadingDock-AutoGenerated-20250115-143205",
  "action": "BLOCK",
  "enabled": true,
  "sourceZones": {
    "objects": [{"type": "SecurityZone", "id": "...", "name": "VLAN-150-LoadingDock"}]
  },
  "destinationNetworks": {
    "objects": [{"type": "Network", "id": "...", "name": "Corporate-Network"}]
  },
  "metadata": {
    "reason": "Perimeter intrusion detected by edge AI - Camera 47",
    "duration": 1800,
    "createdBy": "edge-ai-automation"
  }
}

Response:

{
  "id": "005056BB-0B24-0ed3-0000-004294967400",
  "name": "BLOCK-LoadingDock-AutoGenerated-20250115-143205",
  "action": "BLOCK",
  "enabled": true,
  "metadata": {
    "timestamp": "2025-01-15T14:32:05.225Z",
    "expiryTimestamp": "2025-01-15T15:02:05.225Z"
  }
}


Complete Integration Summary

Integration Point Protocol Latency Bandwidth Purpose
ISE pxGrid WebSocket/HTTPS 50ms 5 Mbps Badge event correlation (authorized vs. unauthorized)
Splunk MLTK HTTPS REST 100ms 15 Mbps Historical pattern validation (anomaly detection)
ThousandEyes HTTPS REST 80ms 10 Mbps Network path health (rule out packet loss false positives)
AppDynamics HTTPS REST 90ms 10 Mbps Application health (rule out corrupt stream false positives)
FTD Firewall HTTPS REST 50ms 2 Mbps Automated network blocking (contain threat)
XDR SecureX HTTPS REST 40ms 2 Mbps Security incident correlation (enrich with AMP, Umbrella, ISE)
ServiceNow HTTPS REST 60ms 3 Mbps Incident ticketing (audit trail, supervisor review)
Webex Teams HTTPS REST 80ms 2 Mbps Mobile supervisor alerts (push notifications)
BMS Honeywell HTTPS REST 500ms 1 Mbps HVAC/lighting control (UC2 building automation)
TOTAL Max 100ms 50 Mbps Multi-source validation + automated response

END OF INTEGRATION ARCHITECTURE

This document provides the complete integration architecture showing: ✓ Physical layer: 120 cameras with PoE+ power and network connectivity ✓ Access layer: Catalyst 9300 switches with SGACL enforcement ✓ Distribution layer: Catalyst 9500 inter-VLAN routing ✓ Edge AI layer: UCS XE9305 + XE130c M8 with NVIDIA L4 GPU inference ✓ Observability layer: ISE, Splunk, ThousandEyes, AppDynamics validation ✓ Security layer: FTD, XDR, ServiceNow, Webex automated response ✓ Complete data flow: 0ms → 500ms detection to mobile alert

Total Bandwidth: 960 Mbps inbound (cameras) + 50 Mbps outbound (APIs) = 1,010 Mbps Network Utilization: 5% of 20 Gbps available capacity (ample headroom for Phase 5)