Request a quote → Server down · Emergency
★ The manifesto behind our work

The vendor says buy a new one. We say repairable.

When a manufacturer declares a server "end-of-service-life", it is announcing a commercial decision — not a technical verdict on the hardware. Almost always the machine is perfectly repairable, and can stay that way for years. This is the work of RiparazioneServer.com.

The problem

"End-of-life" is a business decision. Not a physical fact.

OEM manufacturers work to product cycles. Every 5-7 years they declare a system out of support: they stop selling official service for it, run down spare parts stock and eventually drop those parts from their price lists. From their point of view it is rational — they concentrate resources on new models and push the cyclical upgrade.

But a server engineered for a 10-year service life does not stop working in year seven. The parts that wear out (fans, batteries, thermal paste, spinning disks) get replaced. Electrical components that are healthy stay healthy. A 2018 motherboard with healthy VRMs has no idea it has been declared EOL.

What this means in practice for the customer: faced with a fault on an EOL server, the vendor's typical answer is "buy new" — offering a new system that can cost 10-20 times the value of the repair actually needed.

Our answer

Repair at component level. Source parts outside the channel. Document everything.

We are not an OEM channel. We do not sell cyclical upgrades. Our business model is exactly the opposite: we make your server last as long as possible, within reason.

  • We replace the single failed component, not the whole system.
  • We source spare parts through three channels: official refurbished, certified donors, compatible equivalents (always stated in the quote).
  • If the repair makes no economic sense, we say so — there is no incentive for us to push a pointless job.

The TCO gap is often dramatic. And in many cases it is also the more sustainable choice: a repaired server is a server that does not become electronic waste.

Spare parts procurement

Three channels. Always stated before the job.

Transparency about where a spare part comes from is an absolute rule for us. The customer knows before the job whether the part we are fitting is official, official refurbished or a compatible equivalent — and what commercial warranty follows from that.

Channel 1

Official hardware

Where available, new spare parts straight from the manufacturer's channels. This is the first choice if the system is still in the catalogue or if the vendor keeps parts stock even beyond official EOL.

Typical warranty: 12 months, vendor.

Official channel
Channel 2

Certified refurbished

Components refurbished by qualified suppliers, tested and certified. Often the best value-for-money option on EOL servers. The parts typically originate from controlled decommissioning.

Typical warranty: 12 months on the component.

Refurbished parts
Channel 3

Donor / compatible equivalents

For particularly old models or discontinued brands: identification of donors that are compatible by electrical and mechanical equivalence, or cross-referencing to generic components where applicable (e.g. CPUs on shared sockets).

Typical warranty: 6 months on the component, assessed case by case.

EOL sourcing

Indicative TCO example

Previous-generation PowerEdge server, declared EOL, failed motherboard + 2 DIMMs:

Equivalent new system ....................... € 14.000 – 22.000 OS + application migration .................. € 2.500 – 5.000 Locked software re-licensing ................ € 1.000 – 4.000 Production downtime (2-5 days) .............. variable ──────────────── Indicative TOTAL ............................ ≈ € 17.500 – 31.000
Certified refurbished motherboard ........... € 700 – 1.400 2x equivalent ECC DIMM ...................... € 160 – 320 Labour (diagnosis + repair + test) .......... € 600 – 1.200 Production downtime (1-2 days) .............. limited ──────────────── Indicative TOTAL ............................ ≈ € 1.460 – 2.920

Indicative figures, subject to case-by-case verification. They are not a quote: the real assessment is the quote you request.

When repairing makes sense

Not always. But more often than the vendor thinks.

Repairing is worth it when:

  • The current workload is below the server's capacity (it is not asking for more compute)
  • Applications are certified on the existing OS/hardware and migrating costs time
  • Software licensing is tied to the service tag or to the current system
  • The surrounding infrastructure (storage, network, integrations) is consistent with the system
  • You need to buy 2-3 years to plan the future migration properly

Replacing is worth it when:

  • The old system's annual power draw costs more than the capex difference
  • You need modern security features (TPM 2.0, recent microcode, modern Secure Boot)
  • You want to consolidate several servers into one (GPU density, NVMe)
  • The current system is past the point where spare parts are realistically obtainable

In both cases, an honest diagnosis helps you decide. There is no outcome that "suits us".

Real cases · anonymised

"Dead" servers put back into production.

EOL · declared no longer supported

Accountancy firm, application server

Previous-generation PowerEdge, motherboard VRM failure. Vendor: new system or cloud service. Our assessment: certified refurbished motherboard + full hardware refresh. System back in production, client management software untouched, expected coverage for another 2-3 years to plan the cloud migration calmly.

Brand out of the market

Manufacturing SME, production server

Server from a brand no longer present on the Italian market, two faults: PSU + main CPU fan. Parts sourced through the compatible-equivalent channel, full repair carried out. The production line that depended on the system was running again in 5 working days, versus weeks of downtime for a migration.

8 years · "not worth it"

Non-profit organisation, file server

8-year-old server, failed motherboard. Limited budget; losing the system would have forced a slow rebuild. Repaired with a compatible donor + two new disks (the previous ones already at end of life); the system carries on serving the mission with no further capex spend.

FAQ · beyond the vendor

Questions from people weighing up whether to really "buy new".

The manufacturer says the server can no longer be repaired. Is that true?

Almost never in technical terms. When a vendor declares a system "end-of-service-life", it is saying it will no longer sell official service for it, not that the hardware has become impossible to repair. The components remain repairable and spare parts are often obtainable — just no longer through the manufacturer's official channel. Our experience is that the large majority of servers declared "dead by the vendor" are perfectly recoverable at costs far below a replacement.

Do you source spare parts for servers from brands that have left the market?

Yes, for the more widespread models. For servers from discontinued brands (e.g. past-generation Acer Altos lines, certain out-of-production Fujitsu lines) we work through three channels: certified refurbished from the qualified secondary market, compatible donors identified by electrical/mechanical equivalence, and in some cases cross-referencing to generic components (e.g. CPUs on sockets shared across brands). When a part is not realistically obtainable within acceptable timescales, we tell you up front — not after stripping the server down.

Is it better to repair an EOL server or buy a new one?

It depends on the realistic TCO over 24-36 months. Repairing is worth it when: the workload does not demand more performance than it has now, the applications are certified on the existing OS/hardware, and migrating to a new system would carry hidden costs (re-licensing, rebuilding integrations, downtime). Replacing is worth it when: the old system's power draw is much higher, recent security features are missing (TPM 2.0, up-to-date microcode), or you want to consolidate several servers into one. We always give the honest analysis — even when it says repairing makes no sense.

What warranty do you give on parts fitted to out-of-support servers?

On the replaced component: a working-order warranty of normally 12 months when the part is certified refurbished or new official. For donors from the non-certified secondary market, a warranty proportionate to the nature of the part (typically 6 months). On the labour: a warranty on the work carried out. Our transparency is the rule: before the job you know exactly what kind of part we will fit, where it comes from and what coverage we give you.

Can an EOL server you have spent money on still break tomorrow?

Yes, like any server. What we do is not "guarantee it will never break again", but bring the system back to nominal condition for its original design. Example: an 8-year-old PowerEdge with a new motherboard, healthy ECC RAM, fresh thermal paste and recent disks has a life expectancy equal to that of a similar system that has been in production for 0 years — the vendor's legal warranty is a different thing from technical repairability. For high-criticality workloads we always recommend pairing it with a robust backup (and, for ongoing systems administration, AssistenzaServer.eu).

Been told "it can't be done"?

Send us the model, year and symptoms. We'll tell you if that's true.

Request a quote. If the repair makes no sense, we say so. If it does, the quote separates the part (and which channel it comes from) from the labour.