Cables are a special case. Non-serialized, fungible, high-churn. SSDs have serial numbers, SMART data, finite wear. Different error regime.
Revised estimate: for serialized assets with barcode scan, spreadsheet error rate drops to 5-8% per my audit at last job. Still unacceptable for critical path, but "40% on everything" is misleading framing.
Raspberry Pi monitor: valid. I withdraw objection for <50 SKUs.
I went and checked my "closet full of drives" after this thread. 14 drives, 3 are dead, I never marked them. Spreadsheet said 12 good, reality 11 good. Not 40% but not zero.
Amazon same-day is a lie in Phoenix summer, warehouse workers are human, I get it next day usually. I am switching to a simple Python script on an old laptop, reads a CSV, emails me when count < 2. Took 20 minutes. Fail1996 inspired me, thanks.
Member
Manchester here. I use eBuyer and Scan for next-day, or walk to their depot if desperate. For inventory I use Snipe-IT, self-hosted, overkill for some but I like the audit trail. Asset tags, the whole thing.
Single-location DC with no staff rotation? Maybe overkill. But I inherited 200 mystery boxes labeled "2018" and "???" so I am traumatized. Never again.
In Berlin I use Reichelt and Conrad, both have will-call if I drive 30 minutes. For inventory I tried PartKeepr but it died, project abandoned. Now I use Home Assistant with a custom integration, scan barcode with phone, updates count. Weird but works.
Germany: no same-day culture. 2-3 day is standard. I keep 1 month of spares for anything that would take >3 days to source. That is my rule.
I boot everything from network so local disk failure is "reboot, wait 90 seconds, continue working." For the few boxes that need local boot, I keep 1 spare. Inventory is a text file on my PXE server, edited with vim. Last modified 2024-03. I should check that.
Network boot believer because I saw too many "spare" drives that were already dead. If you do not test your spare, you have a decorative paperweight.
This. I bought 10 "spare" drives from a deal, 4 DOA, 2 failed within month. My $1/GB/yr rule does not apply to spares, I pay more for known-good from B&H or Adorama. Miami = no local tech suppliers, everything ships from NJ or CA.
I use a Google Sheet with conditional formatting, red when last test >90 days. Not automated but visual. I test quarterly, write date on drive with sharpie. Low tech, works.
I am in Bangalore. Local market SP Road has everything same-day but 50% is fake or refurbished sold as new. I learned to check firmware versions. Inventory is mental, I know every shopkeeper by name. Not scalable, not recommendable, but it is what I have.
For real enterprise stuff I use Dell India, 2 week lead time. Keep 2 spares minimum, hide them from my own team or they "borrow" permanently.
Member
Zurich perspective: everything is expensive, everything is slow. I use Digitec for consumer, Brack for pro, both next-day if you pay. I do not keep large spares inventory, I keep IPv4 addresses that appreciate faster than SSDs depreciate. Different risk model.
For the actual thread question: I use a self-hosted GLPI. Overkill? Yes. But Swiss compliance documentation requires traceability, so I have traceability.
I am in Istanbul. Same-day exists for phones, not for server parts. I use Hepsiburada and Trendyol for cables, small things. For disks I have a contact in Germany who ships DHL, 3-4 days.
My inventory system: I take photo of each spare with phone, upload to private Telegram channel, search by scrolling. It is bad. This thread made me feel better about myself, thank you all.
Following this thread. I am in Nairobi, no local server parts market to speak of. Everything comes from Dubai or China, 2-3 weeks. I keep 3 month buffer, cannot afford downtime. Spreadsheet, no automation, I check weekly like a ritual.
Chen your 4 hour drive sounds like luxury to me. At least you have a city.
