N+1 and 2N redundancy levels
Notation stating how many backup units a data center adds to those strictly needed: one extra unit (N+1), or a second complete path (2N).
N is the number of units, generators, UPS modules or chillers, needed to carry the site's load. With N+1, one extra shared unit is installed: the site tolerates the loss of one unit, but not two, and not the failure of a shared element such as a switchboard or a control system. With 2N, the whole path is duplicated, each piece of IT equipment being fed by two independent paths; the loss of an entire path is absorbed, at the price of an installation twice as expensive and a lower load factor. 2N+1 adds a spare unit to each of the two paths, to stay protected during maintenance. The notation describes theoretical capacity, not actual availability: a spare unit that fails to restart, or that no one can restart in time, does not count. Insurers therefore have reason to ask how the redundancy has been tested under real conditions, and what staff is on site at night to make it work.
On August 30, 2023, a voltage sag on the grid tripped the chillers at several Microsoft data centers in the Australia East region. The two affected data halls had seven chillers, five running and two on standby, but only one of the standby units could be started; Microsoft's incident review noted that the three-person night shift was too small to restart the machines in time, and some servers had to be shut down.
Follow the research
One letter a week. First, the week's underwriting note, when one is published: a dated, argued decision, signed by Paul Bernardski. Then what the observatory published on cyber risk, political risk and insurance. No account, no tracking.
No spam, one-click unsubscribe, no data ever sold.