news.volyx.in

Don't rent the cloud, own instead (blog.comma.ai)

1222 points by Torq_boi · 166 days ago · 497 comments on HN

Article summary

The article discusses how the company Comma.ai owns and operates its own data center, which has been running for years and hosts all of their model training, metrics, and data. The data center is maintained by a couple of engineers and technicians and has a total cost of around $5M, which is estimated to be cheaper than renting in the cloud. The company uses a combination of self-built machines, Dell storage, and custom software to manage their compute and storage needs. The data center is designed to be efficient and cost-effective, with features such as outside air cooling and a custom power system.

Main themes

  • Data Center Ownership
  • Cloud Computing
  • Cost Efficiency
  • Infrastructure Management
  • Hardware Customization
  • Disaster Recovery

What commenters say

  • Owning a data center can be more cost-effective than renting in the cloud, especially for companies with consistent workloads.
  • Colocation services can be a viable option for companies that want to own their hardware but not manage a data center.
  • The total cost of ownership of a data center includes not only the initial investment but also ongoing expenses such as maintenance, electricity, and personnel.
  • Some commenters disagree that owning a data center is always the best option, citing the benefits of cloud computing such as scalability and access to the latest hardware.
  • The risk of disasters such as fires or floods is a concern for data center owners, and having a contingency plan in place is crucial.
  • The cost of migrating data out of a cloud provider can be a significant factor in the total cost of ownership calculation.
  • Some companies may prefer to own their hardware and manage their own data center for security and control reasons, rather than relying on a cloud provider.
  • The break-even point for owning a data center versus renting in the cloud can vary depending on factors such as the type of hardware and the location of the data center.