01
Client Profile & Background
A multinational manufacturing company headquartered in Jakarta, operating three production plants in Bekasi, Cikarang, and Karawang, each with its own local data center. The company produces automotive components and relies on real-time inventory management, ERP (SAP), and MES (Manufacturing Execution Systems) running on VMware vSphere clusters. The IT environment includes 200+ virtual machines across 8 hypervisors, with storage area networks (SAN) from Dell EMC and backup appliances from Synology. The network was a legacy three-tier design with Cisco Catalyst switches, causing east-west traffic bottlenecks and manual ACL management.
The company had recently adopted a hybrid cloud strategy using Microsoft Azure for disaster recovery and analytics, but the on-premises network could not efficiently support burst traffic to the cloud. Latency-sensitive MES applications experienced jitter during peak production hours, and security audits revealed inconsistent firewall rules between plants. The CIO mandated a network transformation to support Industry 4.0 initiatives, including predictive maintenance and AI-based quality control.
02
Technical Challenge
The primary challenge was the inability to scale east-west bandwidth: the existing core-distribution-access design limited inter-server throughput to 10 Gbps, while VM migrations and backup traffic required 40 Gbps. During nightly backup windows, the network experienced 30% packet loss, causing backup failures for 12 hours straight. Additionally, the lack of network segmentation meant that a compromised IoT sensor in one plant could laterally move to the ERP system, posing a severe security risk. The IT team spent 20 hours per week manually updating ACLs on 15 switches, with a 5% error rate leading to outages.
Latency was another critical issue: MES transactions between Cikarang and Bekasi plants took 15 ms, exceeding the 5 ms threshold for real-time production monitoring. The company also needed to support 500+ IoT devices collecting vibration and temperature data, which generated 2 TB of daily data that had to be processed locally. The existing network could not handle the multicast traffic from these sensors without flooding the entire VLAN. Finally, the backup infrastructure using Veeam was saturating the network, causing production traffic degradation.
03
Implemented Solution
Intilogy designed and deployed a leaf-spine network fabric using Cisco Nexus 9000 series switches with ACI (Application Centric Infrastructure) for policy-based automation. The fabric consists of 4 spine switches (Nexus 9332C) and 16 leaf switches (Nexus 93180YC-FX3) across the three plants, interconnected via 100 Gbps links. Each leaf connects to servers using 25 Gbps, providing 2:1 oversubscription. The fabric supports VXLAN overlay with EVPN control plane for seamless VM mobility and micro-segmentation. VMware NSX-T was integrated with ACI via a REST API to provide consistent security policies across virtual and physical workloads. For backup, Veeam Backup & Replication v12 was configured with network acceleration and WAN optimization to reduce backup window from 12 hours to 2 hours. Backup & Disaster Recovery was enhanced with Synology Active Backup for Business to offload local backups to a dedicated storage network. The fabric also includes Fortinet FortiGate 600F firewalls at the edge for north-south traffic inspection, integrated with ACI for automated threat response.
04
Results & ROI
Post-implementation, east-west throughput increased to 400 Gbps, eliminating packet loss and reducing backup failures to zero. The backup window shrank from 12 hours to 2 hours, and RPO decreased from 24 hours to 1 hour. Latency between plants dropped from 15 ms to 2 ms, enabling real-time MES synchronization. Security compliance improved with zero-trust micro-segmentation: the attack surface reduced by 90% as lateral movement was blocked by ACI contracts. The IT team saved 20 hours per week on network changes, now automated via ACI APIC. The company achieved a 40% reduction in network TCO over 3 years by eliminating manual processes and reducing downtime. The fabric also supported a 300% increase in IoT devices without performance degradation. Overall, the solution enabled the company to launch a predictive maintenance platform within 6 months, contributing to a 15% increase in production efficiency.
