IBM Cloud’s elasticity stretches and stretches (Big Blue's credibility?)
Slow virtual server provisioning incident mistakenly given Red Alert status
IBM’s cloud is having a bad day.
An incident that kicked off at 19:48 UTC on Thursday, July 5th, saw “long provisioning times [for virtual servers] … observed through monitoring systems,” leading to a warning to the effect that “At this time, customers may experience longer than normal times for provisioning new services.”
Provisioning attempts for new virtual servers (VSI) did work, eventually, and IBM reported “No disruptions have been observed.”
Sometime before 22:10, IBM’s teams heard reports of “VSI provisions stalling”. That news meant “the incident severity has been upgraded to a Sev 1.”
But 20 minutes later IBM told customers that “Information regarding extended provisioning times was incorrectly associated with this incident. Lowering severity back to 2.”
IBM’s most recent update, issued at 02:19 UTC on Friday July 6th advised “We are currently seeing improvements in provisioning times, but this is being attributed to a lower volume due to normal trends. Provisioning transactions continue to be delayed.”
“As of this time there have been no observed provisioning failures associated with this incident. Engineers are continuing to investigate a potential communication issue with the database.”
Server provisioning times has been an IBM Cloud weakness for a while now: The Register’s virtualization desk has tried it and was surprised by several hours delay between clicking “make me this server” and said server’s availability. Rival clouds go far faster for common instance types.
With rapid scaling one of cloud’s premier virtues, a provisioning brownout is not welcome. And given IBM is trying hard to grow its cloud so it rivals market leaders, incorrect assessment of an incident's severity will hardly gladden would-be users, especially as it follows a recent storage wobble and anti-virus service issues. ®