Ein Rechenzentrum in Chicago war überhitzt – und der Handel in wichtigen Märkten auf der ganzen Welt wurde eingestellt
https://www.wsj.com/finance/cme-options-futures-trading-halted-amid-data-center-issue-16e96ed1?st=oFWVdo
Ein Rechenzentrum in Chicago war überhitzt – und der Handel in wichtigen Märkten auf der ganzen Welt wurde eingestellt
https://www.wsj.com/finance/cme-options-futures-trading-halted-amid-data-center-issue-16e96ed1?st=oFWVdo
9 Kommentare
Gift link. Excerpt:
> Inside a nondescript, boxy building in a Chicago suburb, the heat was rising. This wasn’t just an unusual issue for a chilly fall day in Illinois. It was also a problem for global markets.
> The facility was a vital data center used by CME Group, the exchange operator behind some of the most heavily traded instruments in international finance—futures tied to U.S. stock indexes, Treasurys, crude oil and more. And temperatures inside the building were soaring toward 120 degrees Fahrenheit.
> The result was an outage lasting more than 10 hours, slamming the door on trading in these key contracts from late on Thanksgiving evening through to an hour before Wall Street trading opened on Friday morning. While it didn’t stop buying and selling of underlying bonds and stocks, it limited investors’ room to maneuver and made it harder to gauge how U.S. markets would move in the final session of the month.
> In some ways, it was a lucky break, coming during a slow holiday period in the U.S., and mostly resolving before American markets opened. It was also another warning about vulnerabilities in the digital economy: There were echoes of the recent outages at Amazon Web Services and Cloudflare that took down swaths of the internet. The disruptions show what can go wrong when a widely used service that businesses and consumers expect to function around the clock, hits even an apparently mundane technical problem.
Absolute madness to exercise unplanned bcp event when hardware started to melt at 140F
All data centers have huge cooling units to maintain a stable 65°F. But if the cooling system has a problem, temperatures can quick shoot up to 140°F and higher. This is why you have backup data centers where services to be transferred to while you shutdown the servers in the overheating data center. I’d be surprised if that didn’t happen in this case (TLDR).
>the disruption, which it said stemmed from “a chiller plant failure affecting multiple cooling units.”
As a chiller mechanic I can speak to this:
Buildings are built with redundancy:
Multiple chillers, pumps, towers etc
The design concept allows a failure to occur in one component and the redundancy will be there to take over.
This allows repairs to be performed during normal business hours in a non emergency fashion.
What was described above will only result from mismanagement of the facility, where a breakdown is ignored because then backup is working.
When the backup fails as well then you have an emergency as described.
But to say it was “a disruption affecting multiple chillers” is almost impossible short of a power failure.
What should have been said was, “we failed to maintain and repair the primary units and once the secondary failed we had to bring in temporary cooling”
Our electric grid is not advanced and maintained well enough for the AI data center load, meanwhile our citizens will continue to face rising electricity costs to cover the needs of these centers.
Legalize balcony solar nationwide!
Take a gander at Cyrusone.com, scroll to bottom, click on leadership, question 2nd in line.
This is how we defeat them
I’m not sure how wise it would be to divulge that specific information. I can only help point one to the right direction.