4 ms·
I don't buy it (as a backup philosophy). For starters, there is a difference between wanting to replace your data in the event of a drive failure, and wanting y
by dkokelley 15y ago
I don't buy it (as a backup philosophy). For starters, there is a difference between wanting to replace your data in the event of a drive failure, and wanting your descendants 1,000 years from now to be able to read your data.
In the former case, there is nothing special about the 1,000 lifespan, and on a $/GB level, you would do better to use an online backup or redundant drive solution. Let's assume that in 10 years your drive will fail and you need 1TB of data restored. You would need 218 of these disks (or $1,526 worth of the backup medium). Given the same 10 years, an online solution like Carbonite runs $60/year, or $600 for "unlimited" data over 10 years. Granted, on year 11 you lose all of that data unless you keep paying, but there are other benefits like continual backups (whereas this disk is permanently written with 10 year old data, while the online system backed up your photos from last weekend). Assuming you don't want your data in someone else's hands, let's assume you want to store your data on a local hard drive. Of course you're smart and cautious, so you have a raid backup drive. 1TB external drives run for $160. Let's assume a life span of 5 years on average (I've seen better, and I've seen worse). Not counting the original data on your computer's internal drive, you would need 4 drives to cover 10 years (2 drive in a raid configuration, given 5 year's lifespan). This will cost you $640.
However you slice the near-term backup, I can't see the benefits of a permaDVD.
For long-term, archival purposes, this system might make more sense. I'm not sure what other systems are out there that can estimate a 1,000 year lifespan (at $1.50/GB). Then again, imagine trying to find a DVD player in 1,000 years.
- SkyMarshal 15y ago>For long-term, archival purposes, this system might make more sense. I'm not sure what other systems are out there that can estimate a 1,000 year lifespan (at $1.50/GB). Then again, imagine trying to find a DVD player in 1,000 years. Yeah, the article was a little shortsighted for not recognizing that the use case here is not for avg Joe's, but for cloud providers and enterprise datacenters that want better data archival systems. Avg Joe's are going to be migrating to cloud storage, be it Spotify, Amazon Cloud Player, Google Music, Apple, Dropbox, Spideroak, whatever. Those companies are big potential customers for this, depending on how compelling the M-Disc's cost and replacement rates are vs whatever is currently state-of-the-art for archival. I am not a data archiving expert, but it sounds like there are two threats to data stored on media - intrinsic entropy and extrinsic factors like EM, UV, heat, etc. The test they did sounds like it compressed many years of the latter into 24 hours, but is it possible to simulate the former?