1 · Age 5-8 years
The sweet spot. The system is still covered by obtainable spare parts and still performs for traditional workloads, but fans / thermal paste / consumables have reached the end of their life cycle.
What we do
Hardware repair Server maintenance Maintenance contracts Data recovery Enhancement & upgradeBy manufacturer
Dell PowerEdge HPE ProLiant Lenovo ThinkSystem Fujitsu Primergy Supermicro IBM System x / Power Acer AltosEnhancement · areas
RAM expansion CPU upgrade Storage controller NVMe storage Networking Refurbished
Refurbish · service life extension
Operational refurbish: make an enterprise server last, instead of replacing it.
A 5-7 year old enterprise server that still handles its workload can gain 3-5 years of service life through an operational refurbish: deep cleaning, thermal repaste, preventive replacement of worn components (fans, BMC batteries, end-of-life disks), firmware updates, thermal and electrical validation. Cost: a fraction of a new server. Result: a machine that operates like new.
When to refurbish
The four signs that a server is a candidate.
2 · Thermal throttling in the logs
SEL / IPMI shows thermal throttling events under load. The original thermal paste has degraded, the fans are louder or have worn bearings. A refurbish brings the system back into its nominal thermal range.
3 · Noisy fans / vibration
Degraded fan bearings — constant noise, some RPM variance >5% in the logs. Noisy fans are already in decline and fail progressively. Proactive replacement during the refurbish.
4 · Workload still manageable
If the CPU still handles the load and no newer-generation features are needed, refurbish > replacement. We check the load profile first to validate the choice.
What the refurbish covers
Six standardised areas of work.
1 · Deep cleaning
Complete removal of accumulated dust: CPU heatsinks, RAM heatsinks, air filters, fan cage, power supplies. Compressed air and ESD brushes. Result: airflow restored to as-new condition.
2 · Thermal repaste
Removal of the original thermal paste from the CPU heatsinks, application of quality enterprise thermal paste (Arctic MX-6, Noctua NT-H2, in some cases liquid metal for workstations). Measurable difference of 5-15°C on CPU temperatures under load.
3 · Consumable replacement
BMC/CMOS batteries (average life 4-7 years). Fans with evident RPM variance or noisy bearings. VRM and enterprise SSD thermal paste where applicable. Disks with degrading SMART. Visibly bulging capacitors.
4 · Firmware updated
BIOS, BMC, storage controller, NIC, PSU firmware. Updated to the most recent version that supports the existing hardware. Validated before and after for stability. Attention to security microcode (Spectre, Meltdown, downstream patches).
5 · Thermal validation
Post-refurbish: 2-4 hour memtest + Linpack stress test, CPU/DIMM/VRM/ambient temperature monitoring, optimised fan curves, verification that no thermal event appears in the logs. A clean baseline is handed over to the customer.
6 · Full documentation
Written report: parts replaced (part number, serial), parts cleaned, firmware updated (before/after versions), before/after benchmarks where relevant, before/after photographs of the heatsinks. No invisible work.
Platforms we work on
Enterprise server families we can refurbish.
# Dell PowerEdge
Gen14 (R740/R640) In production · refurbish worthwhile
Gen13 (R730/R630) EOL · refurbish + selective upgrade
Gen12 (R720/R620) EOL · refurbish with feature limits
Gen11 / pre-2012 Spare parts sourcing slower
# HPE ProLiant
Gen11 (DL380/DL360) In production
Gen10 (DL380/DL360) EOL · refurbish worthwhile
Gen9 (DL380/DL360) EOL · still obtainable
Gen8 EOL · refurbish with firmware caveats
# Lenovo ThinkSystem
SR650 V2 / V3 In production
SR650 (V1) EOL · refurbish worthwhile
# IBM System x
x3650 M4/M5 EOL · still manageable
x3650 M3 and earlier Spare parts harder to source
# Supermicro
SYS-1029 / 2029 EOL · spare parts available
A+ Server (various) Standard refurbish
# Cisco UCS
B-series / C-series Refurbish coordinated with UCSM
Refurbish process
Six phases within a maintenance window.
1 · Pre-work audit
Reading SEL/IPMI, SMART on every disk, fan tachometers, recent temperatures, thermal events, BIOS version, firmware versions of every component. A written report is delivered before the work starts, with red/amber/green priorities.
2 · Planning and window
We agree on the list of work (parts to replace, firmware to update, depth of cleaning), the estimated duration (typically 4-8 hours of work per server) and an agreed maintenance window.
3 · Disassembly and cleaning
Server powered down, complete removal of heatsinks, fan cages, disks and filters. Deep cleaning of every disassembled component. Before photos where useful.
4 · Consumable replacement
Installation of new fans, new BMC/CMOS batteries, and new disks where SMART is in decline. Re-application of quality thermal paste.
5 · Firmware and configuration
Updating BIOS, BMC, storage controller and NIC firmware. Verification that restored BIOS settings match the pre-work state. SEL cleared.
6 · Validation and handover
2-4 hour post-refurbish stress test, written thermal baseline, controlled application restart, final report with every action tracked.
Anonymised real case
Refurbish of 3 PowerEdge R730 after 6 years of continuous production.
An SME manufacturing customer in the Lodi area, 3 Dell PowerEdge R730 servers in production since 2019 (bought new). Workload: file server (1), ERP management software + SQL Server database (1), terminal server for 35 users (1). Condition: noisy fans, weekly thermal events in the iDRAC logs, CPU temperatures under load peaking at 88-92°C (nominal range 70-80°C), visible dust on the heatsinks.
Diagnosis: servers in thermal difficulty but electrically healthy. The workload remains manageable for several more years once thermal performance is restored. Refurbish > replacement.
Execution: extended weekend window (start Saturday morning, first server back in production Saturday evening, second on Sunday, third on Monday night). For each server: heatsink disassembly, deep cleaning (the fan cage held more than 6 years of dust), thermal repaste with Arctic MX-6, replacement of all 8 fans (2 with noisy bearings, the other 6 preventively), new BMC battery, BIOS and iDRAC updated to a recent version.
Result: CPU temperatures under load down to 64-72°C on every node (-20°C), zero thermal events in the following months, drastically reduced noise, a system that operates like new. Service life extended by an estimated 4-5 additional years — refurbish cost roughly 15-18% of the cost of full replacement.
# Before · R730 · 6 years in production
CPU temp peak 88-92°C
CPU temp avg 76-82°C
Thermal events 3-5/week
FAN RPM variance 8-12% (some)
Throttling documented in the logs
# After · R730 refurbish
CPU temp peak 64-72°C
CPU temp avg 55-62°C
Thermal events 0 (4 weeks)
FAN RPM variance <2% new fans
Throttling none
# Components replaced per node
8 fans · 1 BMC battery · thermal paste
firmware: iDRAC 2.x → 4.x · BIOS 2.x → 2.x recent
Cost drivers
Four line items on a refurbish quote.
- Components replaced — fans (as a batch), batteries, enterprise thermal paste, any end-of-life disks. The most variable line item.
- Labour (4-8 hours per server) — deep disassembly, cleaning, repaste, firmware, validation.
- Maintenance window — overnight or weekend, on single systems or small clusters. Cost is proportional to how critical the window is.
- Any upgrades done in parallel — if RAM, CPU or NICs are added during the refurbish, the line items add up. It often makes sense to take advantage of the server being powered down and do all the work at once.
FAQ
The questions we are asked most often.
What is the difference between an operational refurbish and a hardware refresh?
A hardware refresh is scheduled maintenance: cleaning, thermal repaste, replacement of a few standard parts (BMC batteries, possibly a fan). An operational refurbish goes deeper: on top of the refresh, we proactively replace components close to end of life (fans flagged by derived SMART indicators, disks with declining SMART, heatsink upgrades where needed), update firmware across an extended matrix and validate thermals under load. Output: the server is optimised for another life cycle.
How much service life does it add?
Typically 3-5 years of additional operation on a 5-7 year old server. The mechanics are cleaned, the fans are reconditioned (or replaced), the thermal paste becomes effective again, thermal headroom returns to nominal. The electronic subsystem stays what it always was — enterprise servers are designed for 10+ years — but the consumable parts return to the condition of a young system.
Is it worth it compared with buying a new server?
Often yes, if the workload does not benefit from a new CPU generation. The refurbish cost is a fraction of the cost of renewal, there is no migration labour, software licences are unchanged and the network topology is unchanged. It is NOT worth refurbishing if: the CPU can no longer handle the current workload, or features the system lacks are required (native NVMe, 25G+ networking, AVX-512). In those cases we look first at a targeted upgrade (CPU, NIC), then at the refurbish.
Which servers do you refurbish?
Standard enterprise families: Dell PowerEdge (Gen12 onwards), HPE ProLiant (Gen8/Gen9/Gen10/Gen11), Lenovo ThinkSystem, IBM System x, Supermicro, Fujitsu Primergy, Cisco UCS. We also work on vendor-EOL systems (Dell PowerEdge Gen11, HPE ProLiant Gen7, pre-2014 servers), but sourcing spare parts can take longer.
Do you give a warranty on the refurbish?
Yes, on the components we replace (fans, disks, batteries, heatsinks) there is a minimum 12-month warranty, and 24 months on certified enterprise spare parts. On the complete system we cannot warrant the components we did not replace (they are yours), but we do warrant the documented thermal and functional state after the refurbish. Everything is tracked in the service report.
Are refurbish and refurbished the same thing?
No. Refurbish (the verb/the operation) is what WE do on YOUR servers. A refurbished SERVER (the noun) is a server that has already been reconditioned and is sold as such (through the tnsolutions channel or third parties). This page is about refurbish: work on an existing customer asset. For refurbished servers ready to buy: contact us or see the tnsolutions.it website.
Let's start a conversation
Tell me the brand, the model and the goal. I'll come back with a plan.
Send me the brand, the model (Service Tag / Serial / motherboard part number) and the target workload. Within one working day I'll reply with the technical feasibility, the constraints I've spotted and an honest estimate.