An exabyte of disk storage at CERN | Hacker News Reader

An exabyte of disk storage at CERN

home.cern

55 pointskakokeko24 comments

This is impressive but it's estimated S3 is at least 1 or 2 zettabytes.

https://www.reddit.com/r/dataengineering/comments/y6lb2v/est...

That's about 1/18th of the 64 bit address space!

64 bits should be enough for anybody. I remember reading exabyte many years back and thinking it was a truly absurd number to be in one place

Even storing everything in 100 Gigabyte “8k Blu-ray quality chunks”, you would have to be storing 10 Million Movies and TV shows to reach 1 exabyte worth of storage. At an average of 30 minutes each, that would take something like 500 years to watch all of 24/7.

And that’s 1/18th the total space a 64 bit number could address. So yeah, 64 bits should be enough.

4K Blu-ray quality video requires a 144 Mbps bit rate. On a triple layer Blu-ray Disc (100GB in size) that nets you ~90 minutes of run time. That doesn't include audio, which can require extraordinary amounts of space as well.

For comparable quality 8K recordings you need 4 time as much space as needed for 4K.

8K won't be the stopping point either.

Codec plays a huge role here.

By your calculations the 4k blu-ray box set for Lord of the Rings, Return of the King would take 5 triple layered disks (for both the extended and theatrical versions it comes with) for just the video and no audio or anything else like the behind the scenes content when the box set only comes with 9 disks for all 3 movies. Likely not using the full bitrate you provided but still. Hard to say they aren’t getting really high quality content on the screen.

In case anyone is curious, I took a quick look at my copy of the set:

- The Fellowship of the Ring

    - Theatrical:      66.7 GiB video @ 53.5 Mbps, 3.81 GiB English audio @ 3055 kbps

    - Extended disc 1: 51.5 GiB video @ 69.8 Mbps, 2.22 GiB English audio @ 3000 kbps

    - Extended disc 2: 56.8 GiB video @ 66.4 Mbps, 2.56 GiB English audio @ 2989 kbps

- The Two Towers

    - Theatrical:      72.4 GiB video @ 57.8 Mbps, 3.90 GiB English audio @ 3108 kbps

    - Extended disc 1: 49.8 GiB video @ 66.9 Mbps, 2.30 GiB English audio @ 3089 kbps

    - Extended disc 2: 52.6 GiB video @ 58.4 Mbps, 2.73 GiB English audio @ 3032 kbps

- The Return of the King

    - Theatrical:      73.2 GiB video @ 52.1 Mbps, 4.33 GiB English audio @ 3086 kbps

    - Extended disc 1: 59.2 GiB video @ 66.3 Mbps, 2.66 GiB English audio @ 2985 kbps

    - Extended disc 2: 61.5 GiB video @ 64.9 Mbps, 2.80 GiB English audio @ 2957 kbps

Actually, putting it this way makes me think that 64 bits may not be enough for anybody. 1/18th of a total space is a lot. And as soon as you try to use any of those bits for routing or subdivision of your capacity, you start underutilizing the whole space.

And if you argue that you should keep your routing info out of your addresses, then I feel like that's already an admission that the address isn't large enough.

note: ipv6 is 128 bits, which should, in fact, be enough for everybody

We’re definitely on the verge of it not being enough.

https://www.computerworld.com/article/2960642/cerns-data-sto...

“In the next 10 to 20 years, data will grow immensely because the intensity of accelerator will be ramped up”

https://home.cern/science/computing/storage

“Even after the drastic data reduction performed by the experiments, the CERN Data Centre used to process on average one petabyte (one million gigabytes) of data per day during LHC Run 2.

The LHC experiments plan to collect more data during LHC Run 3 than they did in the first two runs combined.”

Nothing a FreeNAS running ZFS can't handle.

With enough RAM, i suppose.

It's not entirely clear if this is physical space or effective space (available after raid and parity).

> CERN’s data store has now crossed the remarkable capacity threshold of one exabyte, meaning that CERN has one million terabytes of disk space ready for data!

1MTB -> 1EB ready for data. It is _effective space_, not raw space. And it keeps growing.

Ha, I remember doing a[0] 3-month stint as a student there back when a single PiB got oohs-and-aaahs! So crazy to see how much happened in terms of storage capacity in the intervening 20+ years.

[0] Completely insignificant to them, but a good learning experience to me!

"the one terabyte per second (1 TB/s) threshold". This is definitely something extremely good for the research pipelines. Also, Moore's Law and Kryder's Law seems not applicable that much on their current state.

I thought CERN was running straight CEPH. I was incorrect.

https://eos-web.web.cern.ch/eos-web/

How would you even fill that? Particle collisions might generate a lot of data, but I find it hard to imagine it would need that much

Installing Ark: Survival Evolved will just about get you there

Speaking as someone who designs hardware to process data streaming off just one of the detectors on one of the experiments at the LHC, its hard to imagine that is enough!

As a computational physicist, I often hear people in industry talk about "big data" and then laugh when I understand the amount of data they're talking about. I once accidentally wrote a hundred TiB to tape as part of a side project.

CERN is on a whole other level.

"Big data" used to be thrown around a lot. It's not "big data" if it could fit on any kind of single PC available commercially.

https://atlas.cern/Discover/Detector/Trigger-DAQ

The raw LHC data is already filtered down to a tiny percentage (<0.1% IIRC) using several levels of hardware and then software triggers to make sure only the interesting stuff is recorded. So LHC could very easily fill several orders of magnitude more than this.

Reply on news.ycombinator.com