Hacker News new | ask | show | jobs
by rs23296008n1 2176 days ago
More is better. Those RAIDs are never satisfied.

Its a pity we don't have anything remotely similar for optical storage. Physical limits and all that.

2 comments

More is not better. Repair time gets slower as the ratio for transport speed to capacity gets lower and lower. This means that more file pieces can be lost during the longer repair window resulting in possible data loss.

Keep in mind that when people calculate lifetimes and repairs they usually assume that drive failures are independent. Given that they are installed at the same time and usually from the same manufacturer and bought at the same time (e.g. same batch), it's not great to assume that all these things are independent.

In my 50 TB RAID 5, it has never finished a monthly scheduled raid sync because it takes roughly a week to complete, and performance during the sync is so bad that it is unusable.
You can always buy lower capacity drives and plug more in.
Yeah, what happened to all those 3d optical storage startups that promised us petabyte sized things by now?
Possibilities :

Worked & couldn't gain traction. Capital requirements too great. Eg just a lack of funding. Limited prototypes produced but not mature enough. Tech is quietly smouldering while it is passed from company to company. Slow progress.

Worked & got absorbed. Now in some black ops lab quietly still in development with protoypes already in limited use. A limited commercial version is 5, 10 or 20 years away. The restricted version in use within 2-5 years.

Worked & still in commercial development. Fundamental issues are blocking major progress. Possible confusion over usage priorities is diluting resources, eg they are optimising for one or two use cases and those aren't particularly helping guide development, yet those use cases are also where funding is available. 10-30 years.

Don't work & too unreliable. R&D couldnt get a stable and reliable product due to fundamental issues.

Don't work & unreliable & not enough cash. The prototypes sort of worked but had issues and on top there wasn't enough capital to iron out the issues in the early prototypes.

Other options also exist. The timing started 10 years ago. Should start to see results, if any, from the "worked" options sometime this or next decade if that happened and to whatever extent. A lazy web search still shows some names. A few have quietly disappeared.

Working for "Government agencies"? (Assuming they haven't gone broke)