Page MenuHomePhabricator

annotations.txt
No OneTemporary

Size
1 KB
Referenced Files
None
Subscribers
None

annotations.txt

AFF sector annotations.
Why? Becuase there are many cases where we want to annotate on a sector-by-sector basis:
* Bad blocks
* Blank sectors
* Accessed sectors.
Functionality we want:
* Set annotation
* Query
* Find sectors with an annotation?
Model: A 1TB disk has 2 billion sectors and 62,500 AFF pages (wow;
what happens to AFF performance when all of that is stored in a
sequential, unsorted list?)
If we store these annotations as a byte array, that's 2GB for the 1TB disk. Not acceptable.
Storing it as bits is 256MB; still pretty harsh.
Store the annotations per page and you have 4096 bytes with a 1-bit annotation.
- That's okay for a bad block list.
for a byte annotation per sector, there are 32K 512-byte sectors on a page.
Another option: Store as a sequence of [start block, run length, annotation]
- Good for sparse, bad for filled.
- So you would need a way to turn the sparse into non-sparse

File Metadata

Mime Type
text/plain
Expires
Fri, Nov 21, 9:55 PM (1 d, 12 h)
Storage Engine
blob
Storage Format
Raw Data
Storage Handle
1361535
Default Alt Text
annotations.txt (1 KB)

Event Timeline