Talk:Amazon S3 Glacier

Underlying hardware used by service
There appears to be a school of thought that this is tape based, not custom HDD based. The entry should be edited to reflect that.


 * Yes, this misconception is common - even among software developers who have used the service. I added a small paragraph about this with a source -- cookingwithrye (talk) 06:17, 5 March 2013 (UTC)

Amazon Glacier is part of Amazon Simple Storage Service
As observed on November 14, 2013, Glacier is just a storage option on S3. Perhaps a historical remark could be made that they were once separate (I suppose), and perhaps also when this changed, but otherwise there seems to be little use in contrasting them as if they were separate services, in this article or in related ones, and maybe even the whole article could be merged into the one about S3, with options Standard, Reduced Redundancy, and Glacier. The article about S3 now even has no information about Standard vs. Reduced Redundancy. RFST (talk) 12:57, 14 November 2013 (UTC)

Typical tape-based data access times
Unlike what is said in the article, tape access time is nowhere close to the 3-5 hours range (and the article referred to in that passage isn't saying that either). A typical tape robot will load a cartridge in less than 20 seconds, a typical LTO drive will thread it in 15 seconds and seek to any block in an average time of 1 minute. So, unless the robot and drives are busy with other tasks and queuing ensues, it takes less than 2 minutes to access the first byte of any file on any tape of a tape library. Of course, busy robotics and drives may sometimes extend this to hours, but that is an exception not the rule. Bgoupil (talk) 10:04, 4 August 2020 (UTC)