Hacker News Reader

Comment by nickjj | Hacker News Reader

One thing to think about with S3 is there's use cases where the price is very low which the article didn't mention.

For example maybe you have ~500 GB of data across millions of objects that has accumulated over 10 years. You don't even know how many reads or writes you have on a monthly basis because your S3 bill is $11 while your total AWS bill is orders of magnitude more.

If you're in a spot like this, moving to R2 to potentially save $7 or whatever it ends up being would end up being a lot more expensive from the engineering costs to do the move. Plus there's old links that might be pointing to a public S3 object which would break if you moved them to another location such as email campaign links, etc..

View on HN

Even simpler: I'm using Glacier Deep Archive for my personal backups, and I don't see how R2 would be cheaper for me.

Have you priced the retrieval cost? You quickly run into high 3 and then 4 figures worth of bandwidth.

Retrieval? For an external backup? If I need to restore and my local backup is completely down, it either means I lost two drives (very unlikely) or the house is a calcinated husk and at this point I'm insured.

And let's be honest. If the house burns down, the computers are the third thing I get out of there after the wife and the dog. My external backup is peace of mind, nothing more. I don't ever expect to need it in my lifetime.

Yes, but if it's your third location of 3-2-1 then it can also make sense to weigh it against data recovery costs on damaged hardware.

I backup to Glacier as well. For me to need to pull from it (and pay that $90/TB or so) means I've lost more than two drives in a historically very reliable RAIDZ2 pool, or lost my NAS entirely.

I'll pay $90/TB over unknown $$$$ for a data recovery from burned/flooded/fried/failed disks.

So you don't test your backups to Glacier ?

Of course. I don't retreive all 4+TB every week or month.

High 3 and 4 figures wouldn't occur for personal backups though. I've done a big retrieval once and the cost was literally just single digits dollars for me. So the total lifetime cost (including retrievals) is cheaper on S3 than R2 for my personal backup use case. This is why I struggle to take seriously any analysis that says S3 is expensive -- it is only expensive if you use the most expensive (default) S3 product. S3 has more options to offer than than R2 or other competitors which is why I stay with S3 and pay <$1.00 a month for my entire backup. Most competitors (including R2) would have me pay significantly more than I spend on the appropriate S3 product.

Curious, did you go through the math of figuring out how much the initial file transfer and ongoing cost will set you back (not a lot from the sounds of it). Should be way to do, but I’ve just not found the time yet to do that for a backup I’m intending to send to S3 as well

Yes- AWS calculator: 1) S3- cost for how much you're storing in glacier and then the cost for its metadata to he in a S3 hot/warm tier. 2) Glacier retrieval cost- can be minimized by letting AWS pull it on their schedule. 3) BANDWIDTH- this is the vast majority of the cost at $0.09/GB. There may be ways to lower this, but I haven't tried them.

I'm not saying its a bad deal, I very much wish I'd gone with deep archive (especially with the recent addition of "update if different" to the API)

Worst case, it's $90/TB of bandwidth, the retrieval fee, and GB-seconds for a hot S3 tier while waiting to he downloaded and then deleted, plus pocket change for API calls.

With 10TB drives being available for $80-ish... I expect a lot of semi savy people to upload tens of terabytes then cry when they see the bill.

Numbers rounded a bit for simplicity.

I think the most reasonable way to analyze this puts non-instant-access Glacier in a separate category from the rest of S3. R2 doesn't beat it, but R2 is not a competitor in the first place.

Reply on news.ycombinator.com