bk99.de entertain the web since 1997

bk99.de
OVHcloud · Administration & operationsCurated

Fire at the OVH data centre in Strasbourg

Summary

In the night to 10 March 2021 a fire broke out at 00:47 at the Strasbourg site of OVHcloud, one of Europe’s largest hosting providers. For firefighting, power to the entire site with four data centres was switched off; according to OVHcloud about 120,000 services were fully or partially affected. The fire became a lesson in the fact that backups at the same location are no backups.

What OVHcloud reported

According to the statement, fire detection triggered immediately and firefighters arrived quickly; however, they could only start extinguishing after the entire site had been de-energised. Mainly the SBG1 and SBG2 data centres were affected. Authorities and insurers were to determine the cause, which according to OVH could take months.

OVHcloud provided customers with replacements in other data centres: 14,472 dedicated servers were delivered there and 30,775 virtual servers restored. Of about 120,000 affected services, 113,000 had been fully restored at the time of the statement. OVH announced free services for affected customers.

Ideas

  • A site is a single point of failure, even with several buildings.
  • Backups belong in a different place from the data.
  • Cutting power for firefighters also hits undamaged systems.
  • Recovery needs spare capacity, not just backups.

Insights

  • Many customers had booked their backups with the same provider at the same site and lost both. The fire made the old 3-2-1 rule topical again overnight.
  • Switching off intact data centres on the site also shows that physical proximity is a shared risk.

Facts

  • The fire broke out on 10 March 2021 at 00:47.
  • For firefighting, the entire site with four data centres was de-energised.
  • About 120,000 services were affected, 113,000 of them restored at the time of the statement.

References

Remarks

  • The linked page is a continuously updated statement by OVHcloud; the figures come from the status published there.

Recommendations

  • Make backups according to the 3-2-1 rule: three copies, two media, one at another location.
  • With booked backups, check which data centre they are really in.
  • Regularly practise restoring, not just backing up.

Read the original statement

Search the Web Archive

HB · 2005

Matching subjective server outages against monitoring data

Manuel Schmitt describes a customer who reports frequent outages, although monitoring shows continuous availability and no fault had been reported before. The operator’s monitoring showed the server as permanently reachable.

MuM · 2026

Why experience protects against AI: three limits of automation

Sascha Friesike, professor and director at the Weizenbaum Institute for the Networked Society, explains on the rbb channel Menschen und Muster why AI will not replace human work in the foreseeable future but will change it. The video draws on three research findings: Polanyi's paradox of tacit knowledge, Lisanne Bainbridge's ironies of automation and the observation that expectations grow with technical possibilities. This post summarises the arguments, checks the cited studies and shows what follows for working with AI in everyday professional life.