Published SLA: 15-min Critical · 1-hr HighPune · Nashik · Satara · Ahilyanagar · Sambhajinagar · Kolhapur
NetSecure Global
  1. Home
  2. Work
  3. Laser fabrication plant
Case study · Precision laser fabrication

Hybrid cloud case study: three sites, one cloud, zero unplanned shop-floor downtime.

How NetSecure Global engineered a 3-site hybrid cloud infrastructure with ERP and an encrypted multi-site mesh for a high-precision CNC laser cutting, bending and heavy fabrication business. Client name withheld by policy.

Client · name withheldScope · 3 sites + AWS + public VPS30+ remote-desktop users · 20 endpoints
99.98%
Uptime
0
Unplanned shop-floor downtime
<1s
Remote-desktop session latency
15min
Offsite DR sync
90%+
Smart Fetch design efficiency increase
Interactive topology

Click any site. See how it's built.

Laser fabrication plant production topology: three connected sites, a cloud region and a public VPS joined by encrypted site-to-site tunnelsEncrypted tunnelEncrypted tunnelsite-to-site meshERP syncEncrypted tunnelWANLeased line + broadbandAWSERP · compute · databasePublic VPSERP replica · web frontSITE 1 · PLANT HQMulti-WAN firewall + IPSVirtualisation platform (2 hosts)Network storage · Smart FetchERP · MES · 30+ remote-desktop usersRemoteEncrypted VPNSite 2 · PlantDual BB · virtual hostsSite 3 · DRBackup storage · offsite
Site 1

Primary production & plant HQ

  • WAN: leased line (primary) + broadband (secondary) with multi-WAN failover
  • Firewall: firewall with traffic shaping, intrusion detection and prevention, and VLAN isolation
  • Virtualization: dual hypervisor: one platform for the directory and remote-desktop servers, a second for the virtual firewall perimeter, on-prem ERP and the web-based production planning tool (MES)
  • Storage: network storage in RAID 5, 12 TB pool, 8 TB active (~66.7%), Smart Fetch performance tier
  • Remote access: encrypted VPN for field engineers

Tip: use Tab and Enter to move between sites.

Storage engineering & Smart Fetch

The network storage runs RAID 5 across four bays: a 12 TB pool carrying an 8 TB active manufacturing footprint, roughly 66.7% utilisation. We monitor against a critical 70–80% threshold for CAD/CAM storage, mitigate RAID 5 rebuild risk, and work around 1GbE bottlenecks with multichannel file sharing.

Smart Fetch, our flash-accelerated storage design, speeds up random IOPS, directory metadata and multi-part CAD assembly parsing. It increases efficiency by 90% and productivity by more than 60%, with the fast tier sized to the 8 TB working set so the mechanical disks are spared almost all random I/O.

  • Eliminates assembly bottlenecks for 20 workstations and 30 remote-desktop sessions
  • Reduces spindle wear and thermal stress on the HDDs
  • SSD endurance and terabytes written (TBW) tracked continuously
  • USB/SNMP UPS integration protects read-write cache integrity during power cuts
  • Metadata pinning for instant folder indexing

Multi-tier backup & disaster recovery

  • Deduplicating backup server: deduplicated, incremental, chunk-based backups of the directory and remote-desktop VMs with near-zero RPO.
  • Image-level backup and DR: protection for every VM on the second host (the virtual firewall, ERP and the production-planning MES), with changed-block tracking and instant VM recovery.
  • Storage to backup host: automated pulls of file shares and CAD repositories onto high-capacity backup storage.
  • Replication: Snapshot replication and versioned backup across the encrypted mesh to Site 3 and AWS object storage.

Applications

ERP (on-prem production instance, with a cloud instance on AWS and a replica on the public VPS) and a proprietary web-based production planning tool (MES).

What we'd harden next

We publish our own critique. For this estate, the roadmap includes:

  • Migrating the RAID 5 pool toward RAID 6 / RAIDZ2 as capacity grows
  • Automated, encrypted offsite configuration backups for every firewall
  • Centralised cross-site telemetry (a SIEM)
  • Decommissioning any legacy server operating systems ahead of end of support
"If a provider can't draw your network, they shouldn't be paid to run it." This is the drawing.

Replicate this infrastructure in your facility.

Start with the same on-site engineering audit.