Seeing the Impacts of Spectre on AWS

Seen some great charts this week from the #Meltdown patch, here’s memcached getting hammered due to RDS latency. pic.twitter.com/lxNQAA4JZ6

— Ray’s condition (@_rayjenkins) January 7, 2018

The SolarWinds AppOptics blog has a great collection of their observed effects from recent Spectre/Meltdown patches to AWS. My favorite one is the slight bump in cache failures embedded above. While SolarWinds eventually saw an overall reduction in the CPU impact, it was still easily observable. As subsequent patches are able to take a more refined approach to the problems of the two bugs, we’ll probably see less of an overwhelming performance difference.

Still, as predicted, while the CPU usage increased up to 25% after the patch, the bigger impact appears to be overall latency. This saw huge increases across the board, from 45-100% depending on the instance and service. While throwing more capacity at the problem is possible, the added latency can have a long tail of issues for many services. While public cloud providers absorb a lot of the headache in the wake of the Spectre/Meltdown patch, it doesn’t mean their customers are entirely unaffected from it.

SolarWinds Comments:

In this post, I’ll highlight the impacts we at SolarWinds Cloud® noticed across our AWS infrastructure during the last several weeks due to Meltdown. We, along with many SaaS companies, were impacted by these changes and suffered partial downtime due to AWS efforts to mitigate Meltdown. We’ll be posting a full postmortem for our incident in the coming days. Although this is the impact we saw in our environment, we realize this may not be the same for other environments.

Read More

• About the Author
• Latest Posts

About Rich Stroffolino

Rich has been a tech enthusiast since he first used the speech simulator on a Magnavox Odyssey². Current areas of interest include ZFS, the false hopes of memristors, and the oral history of Transmeta.

• Phoummala Schmitt – IT Origins - September 20, 2018
• Joining Microsoft’s Teams, or No More Slacking - September 20, 2018
• ARM Servers and End Times | Gestalt IT Rundown: September 19, 2018 - September 19, 2018
• NetApp Acquires StackPointCloud - September 18, 2018
• Networking Disaggregation Isn’t Ready – The On-Premise IT Roundtable - September 18, 2018
• Built for SMBs: A Look at the HPE ProLiant ML110 Gen10 - September 18, 2018
• Pure Storage and the State of VVols - September 14, 2018
• Ken Nalbone – IT Origins - September 13, 2018
• Advanced Persistent Database | Gestalt IT Rundown: September 12, 2018 - September 12, 2018
• How a Storage Company Approaches Containers - September 12, 2018