Live data from Hacker News

Building the heap: racking 30 petabytes of hard drives for pretraining

si.inc

211–220 of 281 posts

Re: Building the heap: racking 30 petabytes of hard drives for pretraining

#213
post #212

Any startup that has enough money to casually buy a two-letter domain name has too much money, period. Kind of like counting the number of Aeron chairs at startups of old. Not a good sign.

It's a .inc domain name, are those worth anything?

Re: Building the heap: racking 30 petabytes of hard drives for pretraining

#214
post #212

Any startup that has enough money to casually buy a two-letter domain name has too much money, period. Kind of like counting the number of Aeron chairs at startups of old. Not a good sign.

Looks like unregistered two letter .inc domains are going for $2300/yr. Certainly <5% of the cost of a single developer.

Re: Building the heap: racking 30 petabytes of hard drives for pretraining

#215
For massive amounts of high-performance storage, the cloud is absolutely the most expensive option, by far. Even just 100+TB is ridiculously expensive on any cloud provider. If your company revolves around large amounts of data, it can make sense to keep it on-prem...

...but only if you compute the TCO. The bandwidth, peering, service contracts, available power, cooling, networking, rack capacity, half-decent smart hands, spare gear, etc, etc. The disks won't be the majority of your bill, and the logistics are difficult. It can still be cheaper than $CLOUD, but you have to deal with all the cost and complexity that comes with DIY, so do your homework first.

Re: Building the heap: racking 30 petabytes of hard drives for pretraining

#216
AWS obviously does the same but better. That's why they are so rich. Their cost is a tiny percentage of their revenue. They buy cheap servers, and then run lots of vms on them. Each of which delivers 10s/100s of $ per month. That server pays for itself in revenue within weeks/months. And it will be in service until it stops working which could be over five years. Same with storage, networking, gpus, etc.

They've spent years optimizing everything so their monthly costs are going to be much lower than what these guys managed on their first attempt. They'll be using less energy. They run their own internet backbones and infrastructure, they design their own hardware and source components directly from the best suppliers, they have exclusive deals with energy providers, etc. Every thing these guys did, AWS does way better. And yet they charge 40x more. AWS at cost price would probably be 60-80x less than what they charge; if not more. Cloudflare undercuts them a bit because they are smaller but they can do the same things. So do MS, Google, and everybody else.

This market is ripe for disruption. There should not be a need to shovel hundreds of billions per year into AWS revenue for the industry. The same business operating at 20% margins would be a game changer. And most of this stuff is commodity stuff. Why is there not more competition in this space driving pricing down aggressively? What's keeping competitors off the market?

Re: Building the heap: racking 30 petabytes of hard drives for pretraining

#218

AWS obviously does the same but better. That's why they are so rich. Their cost is a tiny percentage of their revenue. They buy cheap servers, and then run lots of vms on them. Each of which delivers 10s/100s of $ per month. That server pays for itself in revenue within weeks/months. And it will be in service until it stops working which could be over five years. Same with storage, networking, gpus, etc. They've spen…

Inertia. It's the new 'nobody gets fired for buying..'

A previous project I worked on had relatively little traffic, and AWS costs were rather insane for that.

I proposed exploring OVH or DO and probably get costs down to 2 digits per month. Upper management would hear nothing of it - AWS was what they wanted, costs be damned. They were more protecting their own jobs than making a technical decision, I think.

Post reply on HN