Impossible is Nothing! The 10x hpothesis
summary
Claims measurable 10x gains: Clear Linux goes from multi-minute boot to 400 ms boot, OS updates move to half-second atomic updates, and Caffe optimizations reach 5x now aiming for 10x by changing neural chaining and locking. Containers run inside tiny VMs with KVM tooling at ~150 ms startup and ~20 MB memory, QEMU gets similar fixes, and a prototype cloud (Ciao) launches 10,000 containers plus 5,000 VMs on 100 nodes in about one minute — code on GitHub and demos working.
AI-written summary · about
chapters (9)
- 0:00The 10x Hypothesis
- 0:34Data Center Linux Distribution by Intel
- 1:07Don't settle for 'make boot faster. It's the wrong question to solve! The question to solve is 'make boot fast'.
- 2:24CI/CD Needs Efficient Deployment
- 6:05Symptom of a 10% mindset: Admiring the problem
- 7:58First Challenge: Applying 10x
- 8:52Third Challenge: What About QEMU?
- 10:38Improving OpenStack VM launch
- 12:44Applying Re-invention to Private Cloud
related talks
10mLightning Talk: Millisecond Scale-to-Zero and the Quest to Never Pay for Idle Again
09mLessons Learned from Building Fast Infra for Agents to Access the Internet
15mThe Practice Method to Speed Up 10x Boot-up Time for Guest in Alibaba Cloud
30mBeyond Orchestration: The Cloud Native Runtimes Ecosystem for Performance and Security
20mFrom Cold Start To Warp Speed: Triton Kernel Caching With OCI Container Images
42mWhen the OS gets in the way
08mContainerization – It’s Not Just for CPUs Anymore: Introducin...
24mCut your Cloud Computing Costs by Half with Unikraft