panrobo

*not because Postgres itself is broken, but because the way self-hosted Postgres is consumed is broken

I’m a founder and always get overwhelmed when it comes to finding VCs that fit our product. Over the weekend I built pickYourVC which helps to solve this. Input your startup data and see who is the best VC to lead your next round. For now only for EU founders but if you like it, I will add more data to include US too.

Frontend is done with codex and for backend I used vela, which is the Postgres backend platform that my startup is building. So it was a good way to test our product in prod.

Are you also noticing that Postgres is becoming a default database for AI workloads? Do you think this trend will continue or things will evolve over time?

I guess the abstraction levels over sql might change now with natural language processing capabilities, but I guess there will be sql somehwere down the stack always :D

I feel like these days everyone embraced the fact that SQL is here to stay and it's going to become a standard even for modern AI-driven infrastructures. That's why Postgres is booming and new ecosystem is being created around it. WDYT?

there is a middle ground where you can get NVMe-optimized networked storage with solutions like simplyblock. It gives you local-like performance but separates compute from storage and gives more options for backup & DR.

What do you think of storage space in the context of AI and cloud repatriation? Much happening with upcoming players like VAST, DDN, simplyblock or Hammerspace..

From business perspective it's often hard to see these long-term lock-ins and potential effects of those. And hyperscalers focus in on creating narration where moving to the cloud "accelerates" your business. Sure, you don't need to deal with data center things, but eventually you still have people operating the cloud and it's just different kind of talent - but they are still FTEs you need to manage. So the business question remains on the ROI of the investment and currently data center investments bring very high ROI for those who dare.

actually, the S3 Express One zone - as the name indicates - is not global HA, but single zone. So questioning it over fast block storage like io2 or simplyblock is actually valid. It is just matter of s3 interface over a block interface.

Significant reductions for "fast S3", however the base capacity pricing is still very high at $110/TB/month. That is as expensive as EBS io2 which is much much faster in terms of latency.

aws and other hyperscalers will keep growing, no doubt. Public cloud adoption is at around 20%. So the new companies that migrate into the cloud will keep the growth going. That doesn't deny the fact that some might be repatriating though. Especially ones that couldn't get the benefits out of the cloud.

I don't think the article concludes that companies are ditching the cloud.. :)

does anyone have an idea how s3 can be faster than io2?

Who has tried running Postgres in Kubernetes? What is your take on managed services like RDS vs running self-hosted databases in K8s?

I guess the idea wouldn't be to compete with AWS but just explore potential other revenue streams for Tesla owners. Clearly there will be limitations in terms of what can be done with such computer.

who is also seeing trends for databases to move away from managed services like RDS and move towards k8s?

Discussion about Kubernetes and future of cloud.