Great video man. I've been such a long while trying to understand this by reading books and you solved almost all my doubts in a few minutes. Thank you so much!
About the sector remapping: this happens very rarely and is usually a pre-failure-condition, so if the S.M.A.R.T.-attribute 5 "Reallocated Sector Count" has a raw-value bigger than 0, than the best thing to do is do a last backup and then replace the HDD. Of course it is not always pre-failure, because a damaged sector can also be caused by a bad cable or a power failure (however, in the case of a bad cable, the raw value of the attribute 199 "UltraDMA CRC Error Count" also increases).
Bad cable won't cause a bad sector. The bad data will be coded and written, and the same bad data will be read back successfully (this is why zfs uses end-to-end checksums). Bad power, yes, that MIGHT be able to do it, if ram is corrupted or logic screwed-up, or write-power is insufficient / distorted.
Thank you for good video. I have a qeustion at the last part of video. You said , when we use LBA, hard disk can do sequential access although there is a physical difference between sectors. Is it right?(or, you mean, when we use LBA, hard disk can keep writing if there is a error in 43 section by choosing another sector which is distant from 42,44?) If so, why is it the advantage of LBA? I might not understand the contents in video exactly because I live in non-english country..
The point is that the disk controller may remap block IDs to different physical places, e.g. to hide a failing block. This implies that the HD may handle such an error transparently without the software layers on top having to do anything about it (they won't even notice this). And this is a good thing. Not as good is that the actual access pattern you assume when reading a sequential (logical) sequence of blockIDs may not necessarily be translated to a physical sequential access. Bottom line: As with any indirection in computer science: it gives you some nice features, but you give up some control (which is often crucial in data management).
Carl Smith Notice that standards such as EIDE and ATA-2 constrain the address space that may be used for (C,H,S), e.g. if 28 Bits may be used for the address, C may use 16 bits, H may use 4 bits, and S may use 8 bits. Obviously, if the drive has less net storage space available, not all of these sectors are actually addressable. see also en.wikipedia.org/wiki/Cylinder-head-sector#Heads
Your video is the kind of thing that allows me to put 2+2 together so I can move on. Thanks.
still in 2024 benefiting from you video god bless you
Great video man. I've been such a long while trying to understand this by reading books and you solved almost all my doubts in a few minutes. Thank you so much!
Great explanations. Love the course, it has been a very fun and interesting look at Databases. Thank you!
Really informative video, this has helped me a lot! Thank you for putting this together!
Nice job of expalining a complicated subject. Thanks
thanks!
Thank you for excellent videos
Really clear and helpful video!
Thanks to upload this. It help me a lot.
About the sector remapping: this happens very rarely and is usually a pre-failure-condition, so if the S.M.A.R.T.-attribute 5 "Reallocated Sector Count" has a raw-value bigger than 0, than the best thing to do is do a last backup and then replace the HDD. Of course it is not always pre-failure, because a damaged sector can also be caused by a bad cable or a power failure (however, in the case of a bad cable, the raw value of the attribute 199 "UltraDMA CRC Error Count" also increases).
Bad cable won't cause a bad sector. The bad data will be coded and written, and the same bad data will be read back successfully (this is why zfs uses end-to-end checksums). Bad power, yes, that MIGHT be able to do it, if ram is corrupted or logic screwed-up, or write-power is insufficient / distorted.
Thank you for good video. I have a qeustion at the last part of video. You said , when we use LBA, hard disk can do sequential access although there is a physical difference between sectors. Is it right?(or, you mean, when we use LBA, hard disk can keep writing if there is a error in 43 section by choosing another sector which is distant from 42,44?) If so, why is it the advantage of LBA? I might not understand the contents in video exactly because I live in non-english country..
The point is that the disk controller may remap block IDs to different physical places, e.g. to hide a failing block. This implies that the HD may handle such an error transparently without the software layers on top having to do anything about it (they won't even notice this). And this is a good thing. Not as good is that the actual access pattern you assume when reading a sequential (logical) sequence of blockIDs may not necessarily be translated to a physical sequential access. Bottom line: As with any indirection in computer science: it gives you some nice features, but you give up some control (which is often crucial in data management).
Jens Dittrich Thank you for answer. Ive learned a lot of things from your lectures.
NIce explanation. I wish the lines used for drawing and pointing figures in the video are more legible...
thank you
Great job but IMA still confused about translation of CHS. How can two hard drives with the same CHS be of different size?
Any suggestions on figuring this out...
Carl Smith
Notice that standards such as EIDE and ATA-2 constrain the address space that may be used for (C,H,S), e.g. if 28 Bits may be used for the address, C may use 16 bits, H may use 4 bits, and S may use 8 bits. Obviously, if the drive has less net storage space available, not all of these sectors are actually addressable.
see also en.wikipedia.org/wiki/Cylinder-head-sector#Heads
Thanks for the great explanation.
Super well explained. Thanks.
thanks!
Very clear.
this video very usful for me thanks
This is amazing. thank you!
you are great!
Awesome!!
Ceng'den gelenlere selam olsun :)
Merci beaucoup.
HD picture. Very nice.
Very nice.
thanks!
Do you have anything on SEDs?
still confused by hard disk block and os block
nice...
Dieser deutsche Akzent... D:
Aber danke für die Informationen.