3 ms·

No, that's not right. That's only the data transfer rate. If you check the FAQ, they bill you based on your peak hourly retrieval rate. If I download 4TB at s

by jalada 14y ago

No, that's not right. That's only the data transfer rate. If you check the FAQ, they bill you based on your peak hourly retrieval rate. If I download 4TB at say...10MB/s, not only do I need to pay $480, but I also have to pay ~$257 as a retrieval fee? Their wording is confusing, but ignoring the free retrieval amount (negligible difference on a 4TB transfer): Fee = Peak hourly retrieval * number of hours in month * $0.01 https://aws.amazon.com/glacier/faqs/#How_much_data_can_I_retrieve_for_free https://aws.amazon.com/glacier/faqs/#How_much_data_can_I_ret... Admittedly ~$737 isn't the end of the world if your house has burned down and you need all your data back, but it's still important to know the details. I think in that situation, it would be cheaper to use their bulk import/export, which would be roughly $300 for 4TB

kondro 14y ago
You're right it is complicated. But it also depends on how much you have archived because you get to access 5% of your data on a prorated basis too and that's based on your peak hourly rate. But ultimately, this product isn't designed for backup purposes. It's designed for archive purposes. If you have 4TB of customer data from 3+ years ago that you never access, but need to keep in case the IRS does an audit, then this is the place to put it.
Dylan16807 14y ago
This is so confusing. So apparently if you spend the entire month retrieving the data at 1.6MB/s it only costs $40 plus transfer fees? And more importantly, how do you throttle your retrieval? Edit: So I'm working through a scenario in my head and trying to figure out how charging based on the peak hour isn't completely ridiculous. I have 8GB stored to try out the system. This costs a whopping dollar per year. One day I decide to test out the restore feature. So I go tell Amazon to get my files and wait a few hours. When Amazon is ready, I hit download. I'm on a relatively fast cable connection so the download finishes in an hour. I look at the data transfer prices and expect to be charged one dollar. But I didn't take into account this 'peak hour' method. I just used roughly 8GB/hour over the minimal free retrieval. This gets multiplied out times 24 hours and 30 days to cost 8 * 720 * $0.01 = $57. Fifty-seven times my annual budget because I downloaded my data too quickly after waiting hours for Amazon to get ready.
- freehunter 14y ago
  If you're in a corporate environment, you'd likely have a network admin group that could throttle your connection to them. Or you'd be using a custom-designed front-end that would throttle it. Or if you're SOHO, you could set up QoS rules on your router to throttle it. Realistically though, this service might not be for you if fast and cheap retrieval of your data is important. The importance here is cheap storage, not transfer. They could reasonably expect that you'd only retrieve this data once or twice, if ever, and cost won't be deterrent. Say, if your data center burns down and your company is moving to a new office.
  - Dylan16807 14y ago
    Oh I totally understand that people are going to be willing to pay more after a disaster to get files back. But realistically, if there is a large enough volume of files to bother amazon then they're going to need day(s) to download. If they rate-limited by day then the price would only reach a couple years of storage. The hourly thing is only going to bite minor retrieval events, and it is going to bite them amazingly hard.
    - freehunter 14y ago
      My department (information security) was actually just discussing this service this morning at our morning meeting. We've been looking into backup services for our security monitoring appliance beyond our datacenter and DR site. These backups would need to persist for a year according to PCI/SOX compliance, and if we needed to show data to an auditor, we wouldn't need the entire log. In fact, we'd likely not even need 5% of it. Most likely, we'd only need to pull a day (maybe two) from the logs. I can imagine our media services group talking about the same thing, how to keep master files of their product pictures/videos where they'd only need to grab a file or two here or there (if at all). Amazon seems to be pushing this as a file dump where retrieval of the files is exceedingly rare. They don't want you to use it as a hard drive, they want you to use it as a magnetic tape drive.
      - 14y ago

by jalada 14y ago

kondro 14y ago

You're right it is complicated. But it also depends on how much you have archived because you get to access 5% of your data on a prorated basis too and that's based on your peak hourly rate. But ultimately, this product isn't designed for backup purposes. It's designed for archive purposes. If you have 4TB of customer data from 3+ years ago that you never access, but need to keep in case the IRS does an audit, then this is the place to put it.

Dylan16807 14y ago

This is so confusing. So apparently if you spend the entire month retrieving the data at 1.6MB/s it only costs $40 plus transfer fees? And more importantly, how do you throttle your retrieval? Edit: So I'm working through a scenario in my head and trying to figure out how charging based on the peak hour isn't completely ridiculous. I have 8GB stored to try out the system. This costs a whopping dollar per year. One day I decide to test out the restore feature. So I go tell Amazon to get my files and wait a few hours. When Amazon is ready, I hit download. I'm on a relatively fast cable connection so the download finishes in an hour. I look at the data transfer prices and expect to be charged one dollar. But I didn't take into account this 'peak hour' method. I just used roughly 8GB/hour over the minimal free retrieval. This gets multiplied out times 24 hours and 30 days to cost 8 * 720 * $0.01 = $57. Fifty-seven times my annual budget because I downloaded my data too quickly after waiting hours for Amazon to get ready.

freehunter 14y ago
If you're in a corporate environment, you'd likely have a network admin group that could throttle your connection to them. Or you'd be using a custom-designed front-end that would throttle it. Or if you're SOHO, you could set up QoS rules on your router to throttle it. Realistically though, this service might not be for you if fast and cheap retrieval of your data is important. The importance here is cheap storage, not transfer. They could reasonably expect that you'd only retrieve this data once or twice, if ever, and cost won't be deterrent. Say, if your data center burns down and your company is moving to a new office.
- Dylan16807 14y ago
  Oh I totally understand that people are going to be willing to pay more after a disaster to get files back. But realistically, if there is a large enough volume of files to bother amazon then they're going to need day(s) to download. If they rate-limited by day then the price would only reach a couple years of storage. The hourly thing is only going to bite minor retrieval events, and it is going to bite them amazingly hard.
  - freehunter 14y ago
    My department (information security) was actually just discussing this service this morning at our morning meeting. We've been looking into backup services for our security monitoring appliance beyond our datacenter and DR site. These backups would need to persist for a year according to PCI/SOX compliance, and if we needed to show data to an auditor, we wouldn't need the entire log. In fact, we'd likely not even need 5% of it. Most likely, we'd only need to pull a day (maybe two) from the logs. I can imagine our media services group talking about the same thing, how to keep master files of their product pictures/videos where they'd only need to grab a file or two here or there (if at all). Amazon seems to be pushing this as a file dump where retrieval of the files is exceedingly rare. They don't want you to use it as a hard drive, they want you to use it as a magnetic tape drive.
    - 14y ago