01
I/O Intensive Workloads
Mail, file sharing, and business applications demanding low-latency, high-throughput storage that the SSD/HDD cluster could no longer deliver consistently.
Case Study
Progressive HCI Expansion& NVMe Cluster for High‑IOPS Workloads
Large Non-Profit Organisation — Philippines
As I/O-intensive workloads outgrew the original SSD/HDD HCI cluster, ISPL designed and delivered a new NVMe cluster for high-IOPS VMs — preserving the original investment, avoiding a rip-and-replace, and wrapping the entire two-cluster estate in structured AMC and 24/7 managed services.
Proxmox VEHCINVMeCephProxmox Backup ServerManaged ServicesPhilippines
Scroll
At A Glance
The engagement, in numbers.
2
Production HCI Clusters
NVMe
High-IOPS Storage Tier
14 days
On-Site Migration Sprint
24/7
Managed Services & Helpdesk
Client Environment
An open-source HCI estate managing a growing service portfolio.
The client is a large non-profit organisation based in the Philippines running collaboration, file sharing, and business applications on an open-source hyperconverged infrastructure. The environment consists of two production HCI clusters — both designed and implemented by ISPL: an earlier SSD and HDD-based HCI cluster, and a newer NVMe and SSD-based HCI cluster.
Over time, the client consolidated more and more services onto the HCI platform — mail and collaboration, cloud storage, web services, and line-of-business applications — all running as virtual machines under a unified management plane.
Business Challenge
Storage performance hitting a ceiling — without replacing what works.
As workloads grew and some VMs became highly I/O intensive, the original SSD/HDD cluster alone could not deliver the desired performance headroom and latency for all services while maintaining future growth capacity.
The client's position was clear: scale out the existing architecture — not replace it. Add a new high-performance cluster and selectively move intensive workloads there, while keeping both clusters in active use. Preserve the original investment. Avoid a disruptive rip-and-replace.
02
Growth Capacity Headroom
The original cluster needed performance headroom freed up to handle ongoing growth of standard workloads without becoming a bottleneck.
03
Scale-Out, Not Replace
Replacing the existing cluster would have wasted the original investment and forced a disruptive full migration — unacceptable for a production environment.
04
Unified Operations
Two separate clusters needed coherent management, monitoring, backup, and support coverage — not siloed operations for each environment.
Solution Delivered
A staged HCI programme — original and new clusters forming one logical estate.
Phase 1
Original HCI & Backup Platform
- Commissioned an open-source HCI cluster with SSD and HDD distributed storage based on Ceph, with 10 Gbps switching.
- Migrated existing physical collaboration servers (email, cloud storage, related services) into virtual machines on the cluster — enabling HA, scale-out capacity, and centralized management.
- Integrated an on-site backup node and optional off-site backup capability using Proxmox Backup Server.
Phase 2
New NVMe HCI Cluster for High-IOPS Workloads
- Designed and deployed a NVMe and SSD mixed-storage cluster at the client's central office using modern commodity servers optimized for low-latency storage.
- Planned and executed a logical migration of selected high-IOPS virtual machines — including mail, file, and application workloads — from the original HCI to the new cluster over a ~14-day on-site engagement.
- Integrated the new cluster with existing backup, logging, and monitoring infrastructure for coherent management across both clusters.
Phase 3
Ongoing AMC & Managed Services
- Established a comprehensive AMC and managed services contract covering both clusters and key application VMs, including monitoring, maintenance, security assessment, incident response, and reporting under a detailed task matrix.
- Provided 24/7 helpdesk support from India, with periodic on-site visits.
- Structured capacity planning and failover drills to keep the environment healthy over time.
Architecture Highlights
Two clusters. One coherent, observable estate.
01
Two-Cluster HCI Landscape
An HDD/SSD-based cluster for standard workloads and a three-node NVMe/SSD cluster for high-IOPS VMs — both built on open-source virtualization and distributed storage.
02
High-Speed Networking
Dedicated networks for cluster, storage, and backup traffic — with separate management and data planes for isolation and performance.
03
Proxmox Backup Server Architecture
PBS-based backup with incremental backups, retention policies, and regular restoration drills. On-site primary with optional off-site replication.
04
Unified Observability
Centralised syslog and NMS integration providing unified visibility across both clusters, application VMs, and network infrastructure.
Managed Services & AMC
Structured operations. Documented drills. Continuous coverage.
ISPL established a comprehensive AMC and managed services engagement covering the full two-cluster estate — from hypervisor and storage through applications, security, and distribution network devices.
Infrastructure
- Both HCI clusters
- Hypervisor & storage layer
- Capacity planning & reporting
Applications
- Key application VMs
- Mail, collaboration, file services
- Line-of-business apps
Security
- Security assessments
- Incident response
- Vulnerability management
Operations
- 24/7 helpdesk (India)
- Periodic on-site visits
- Failover drills & documentation
Before vs. After
From performance ceiling to right-sized, right-tiered infrastructure.
Before
- Single SSD/HDD cluster handling all workloads including high-IOPS VMs
- Storage bottlenecks emerging as I/O-intensive workloads grew
- Limited performance headroom — no room for future growth
- Physical servers for collaboration and business apps
- No structured AMC or managed services coverage
After
- Two-cluster HCI landscape — right workload on the right storage tier
- NVMe cluster delivering low-latency, high-IOPS performance for demanding VMs
- Original cluster preserved and extended — original investment protected
- All physical servers virtualized with HA across the estate
- Full AMC and 24/7 managed services — structured ops with documented drills
Outcome
Performance, resilience, scalability — without disruption.
The client now runs a two-cluster HCI landscape where the original platform continues to host suitable workloads, while high-IOPS virtual machines run on the newer NVMe cluster. This progressive expansion avoided a disruptive rip-and-replace approach and preserved the original investment while delivering substantial gains in performance, resilience, and scalability.
Through the AMC and managed services engagement, the client benefits from structured operations, documented failover drills, and proactive capacity planning — all built on an open-source stack that keeps long-term licensing exposure under control.
A living example that infrastructure growth does not require infrastructure replacement — the right architecture accommodates new performance tiers while preserving existing ones.
About Illumia Solutions — As an official Proxmox partner, ISPL designs, deploys, and operates open-source HCI environments for organizations that need enterprise-grade infrastructure without enterprise-grade licensing costs.