| 1 |
Am Montag, 24. November 2008 11:30:25 schrieb William Kenworthy: |
| 2 |
|
| 3 |
> By transient storage I mean that the data is duplicated across across |
| 4 |
> physical storage spaces so that if a machine goes down, the data is |
| 5 |
> still available. |
| 6 |
|
| 7 |
OK, thanks. |
| 8 |
|
| 9 |
> I thought Andrews FS did that, but didnt see when |
| 10 |
> looking at it yesterday. |
| 11 |
|
| 12 |
Yes, (Open-)AFS indeed does this. However, this replication is read-only. This |
| 13 |
means you can read the data as long as at least one replica is available and |
| 14 |
write the data as long as the original (the read-write) volume is available. |
| 15 |
There are also some other things to keep in mind: |
| 16 |
|
| 17 |
* AFS' primary tool for access control are its access control lists (ACL), but |
| 18 |
those are not posix, but AFS ACLs and they apply at the directory (not file) |
| 19 |
level. However, that's usually sufficient, because one can work with subdirs |
| 20 |
and symbolic links to implement more restrictive access for some files in the |
| 21 |
same directory. |
| 22 |
|
| 23 |
* ACLs can also contain host names. |
| 24 |
|
| 25 |
* If a volume is replicated, the client always prefers the read-only path |
| 26 |
(read-write volumes are usually accessed via /afs/.mycell.mydomain, while |
| 27 |
read-only volumes (if they exist) are accessed via /afs/mycell.mydomain). So |
| 28 |
if you want to modify a file you must explicitely open it via the rw-path. |
| 29 |
|
| 30 |
* Replication doesn't happen automatically, needs an explicit command. |
| 31 |
|
| 32 |
* Support for backup volumes is also there (comes with its own backup system). |
| 33 |
|
| 34 |
* Can move volumes to different servers while online. |
| 35 |
|
| 36 |
* Data is cached on the client. |
| 37 |
|
| 38 |
* You'll need Kerberos 5. |
| 39 |
|
| 40 |
If you have further questions, feel free to ask. |
| 41 |
|
| 42 |
Bye... |
| 43 |
|
| 44 |
Dirk |
Archives