Used and refurbished get sold as if they were the same thing. They are not. A used part ships the way it came out of someone else's rack. A refurbished part is inspected, tested, wiped and warrantied by the seller before it ships. The difference is who finds the bad part: the seller's test bench, or your data center.
Four ways hardware reaches the secondary market
- New, factory sealed
- Never opened. Often stock the OEM no longer sells through its own channel.
- Recertified
- From OEM channels such as Cisco Excess and Dell Outlet. Overstock or unused returns that cannot be sold as new.
- Refurbished
- From ITAD channels and vendor networks. Graded, tested, wiped and updated before it ships.
- Used
- Pulled from a decommissioned system and resold as is. Any testing is up to the seller, and often there is none.
What we do before a part ships
Every part goes through a 21-point test, whether it came from an OEM channel or the secondary market. The test runs in four stages.
- Inspection and grading
- Physical inspection for damage and wear. Authenticity checked against OEM specs.
- 48 to 72 hour burn-in
- Full-load stress testing for latent defects. Failures go to repair or recycling.
- OEM-tool diagnostics
- Functional testing with the manufacturer's own diagnostic tools.
- Certification and docs
- Passing parts get certification, documentation and a warranty.
More than 30 percent of the hardware we buy fails this process and never ships.
Why the burn-in matters
Some faults only show under sustained load. A drive, DIMM or power supply can pass a power-on check and fail on day two at full load. The burn-in is where those faults show up. A used part sold as is skips that step, so its first full-load run happens in your rack.
When used is good enough
Used hardware makes sense for a lab, a spare shelf you test yourself, or gear you can afford to lose. For production gear under an SLA, buy tested. If the system it goes into is past its OEM date, check its EOSL date before you buy spares for it. Buying Open Compute hardware? Read what to check on refurbished OCP servers.
Five questions to ask any seller
- Was this exact unit tested, or a sample from the lot?
- How long was the burn-in, and at what load?
- Were the drives wiped, and do I get a record of it?
- Is the firmware updated, and to which version?
- What is the warranty, and who handles a part that arrives dead?
Ask us the same five. Search parts in stock, or price maintenance for the systems they go into.
Short answers
What is the difference between refurbished and used data center hardware?
Used hardware is resold the way it came out of a decommissioned system, with whatever testing the seller chooses to do. Refurbished hardware is inspected, tested, wiped and updated, then warrantied before it ships. At Datacenterservices.ai every part goes through a 21-point test that includes a 48 to 72 hour full-load burn-in.
Is refurbished server hardware safe to run in production?
Test each unit under load before it ships, and it is. A 48 to 72 hour burn-in at full load finds faults a power-on check misses. More than 30 percent of the hardware Datacenterservices.ai buys fails this testing and never ships.
What does recertified mean for data center hardware?
Recertified hardware comes from OEM channels such as Cisco Excess and Dell Outlet. It is overstock or unused returns that cannot be sold as new. Datacenterservices.ai tests recertified parts on the same bench as refurbished ones.