lpx
in [Storage]
|
From: Nik Simpson on 17 Jul 2007 14:24 Arno Wagner wrote: > Previously ikzienietwathierstaat <abcde(a)fgh.ij> wrote: >> thnx, would upgrading to a higher throuput wireless network help, or would >> you not recomend netdisk in wireless networks at all? > > Maybe a really fast would help enopugh, but I doubt it. > It seems the designers of this protocol messed up badly > (a common thing in the industry) by failing to take > realistic networking conditions into account. I would advise > you to stay with cabled connections. They have far, far > superiour properties. > To be fair to the designers, it's making the disk appear locally attached so it has to live with the drive timeout limits of a locally attached hard drive. It's not just LPX, have you tried running iSCSI over a wireless network :-) -- Nik Simpson
From: Arno Wagner on 17 Jul 2007 18:48 Previously Nik Simpson <n_simpson(a)bellsouth.net> wrote: > Arno Wagner wrote: >> Previously ikzienietwathierstaat <abcde(a)fgh.ij> wrote: >>> thnx, would upgrading to a higher throuput wireless network help, or would >>> you not recomend netdisk in wireless networks at all? >> >> Maybe a really fast would help enopugh, but I doubt it. >> It seems the designers of this protocol messed up badly >> (a common thing in the industry) by failing to take >> realistic networking conditions into account. I would advise >> you to stay with cabled connections. They have far, far >> superiour properties. >> > To be fair to the designers, it's making the disk appear locally > attached so it has to live with the drive timeout limits of a locally > attached hard drive. It's not just LPX, have you tried running iSCSI > over a wireless network :-) Hmm. I have not tried that or the ATA sockets Linux offers. But I have used NFS extensively and it does have problems when done with UDP over an unreliable networks. The newer implementations offer NFS over TCP for exactly this reason. Maybe the art of accessing storage over imperfect networks is not that developed after all... Arno
From: Nik Simpson on 17 Jul 2007 20:45 Arno Wagner wrote: > Previously Nik Simpson <n_simpson(a)bellsouth.net> wrote: >> Arno Wagner wrote: >>> Previously ikzienietwathierstaat <abcde(a)fgh.ij> wrote: >>>> thnx, would upgrading to a higher throuput wireless network help, or would >>>> you not recomend netdisk in wireless networks at all? >>> Maybe a really fast would help enopugh, but I doubt it. >>> It seems the designers of this protocol messed up badly >>> (a common thing in the industry) by failing to take >>> realistic networking conditions into account. I would advise >>> you to stay with cabled connections. They have far, far >>> superiour properties. >>> > >> To be fair to the designers, it's making the disk appear locally >> attached so it has to live with the drive timeout limits of a locally >> attached hard drive. It's not just LPX, have you tried running iSCSI >> over a wireless network :-) > > Hmm. I have not tried that or the ATA sockets Linux offers. But I have > used NFS extensively and it does have problems when done with UDP > over an unreliable networks. The newer implementations offer NFS > over TCP for exactly this reason. Maybe the art of accessing storage > over imperfect networks is not that developed after all... > Shared filesystems are completely different beast, what the LPX protocol does is present the storage device as though it was locally attached, i.e. the host it is presented to, sees it as a block device, not a shared filesystem. So it has to beahve like a loccal hard disk, if a local hard disk loses it's connection for even a relatively short period of time the host OS sees a device timeout and marks the drive offline. Network latency over a wireless connection is going to put it right on the edge of timing out pretty much all the time. -- Nik Simpson
From: Arno Wagner on 17 Jul 2007 23:33 Previously Nik Simpson <n_simpson(a)bellsouth.net> wrote: > Arno Wagner wrote: >> Previously Nik Simpson <n_simpson(a)bellsouth.net> wrote: >>> Arno Wagner wrote: >>>> Previously ikzienietwathierstaat <abcde(a)fgh.ij> wrote: >>>>> thnx, would upgrading to a higher throuput wireless network help, or would >>>>> you not recomend netdisk in wireless networks at all? >>>> Maybe a really fast would help enopugh, but I doubt it. >>>> It seems the designers of this protocol messed up badly >>>> (a common thing in the industry) by failing to take >>>> realistic networking conditions into account. I would advise >>>> you to stay with cabled connections. They have far, far >>>> superiour properties. >>>> >> >>> To be fair to the designers, it's making the disk appear locally >>> attached so it has to live with the drive timeout limits of a locally >>> attached hard drive. It's not just LPX, have you tried running iSCSI >>> over a wireless network :-) >> >> Hmm. I have not tried that or the ATA sockets Linux offers. But I have >> used NFS extensively and it does have problems when done with UDP >> over an unreliable networks. The newer implementations offer NFS >> over TCP for exactly this reason. Maybe the art of accessing storage >> over imperfect networks is not that developed after all... >> > Shared filesystems are completely different beast, what the LPX protocol > does is present the storage device as though it was locally attached, > i.e. the host it is presented to, sees it as a block device, not a > shared filesystem. So it has to beahve like a loccal hard disk, if a > local hard disk loses it's connection for even a relatively short period > of time the host OS sees a device timeout and marks the drive offline. > Network latency over a wireless connection is going to put it right on > the edge of timing out pretty much all the time. But this is not necessary. Drive timouts are in the second-range. Otherwise drives would be disabled by the OS all the time. I think the implementers of this just failed to add a decoupling layer, either because they did not want to invest the effort or because they did not undertsand the problem they are solving. Arno
From: ikzienietwathierstaat on 19 Jul 2007 17:00 I am not an expert on this stuff. Would this solve the problem. If so where can I find it (to download). I am running Windows XP SP 1 "Nik Simpson" <n_simpson(a)bellsouth.net> schreef in bericht news:sP7ni.20$yW2.14(a)bignews3.bellsouth.net... > Arno Wagner wrote: > > Previously ikzienietwathierstaat <abcde(a)fgh.ij> wrote: > >> thnx, would upgrading to a higher throuput wireless network help, or would > >> you not recomend netdisk in wireless networks at all? > > > > Maybe a really fast would help enopugh, but I doubt it. > > It seems the designers of this protocol messed up badly > > (a common thing in the industry) by failing to take > > realistic networking conditions into account. I would advise > > you to stay with cabled connections. They have far, far > > superiour properties. > > > > To be fair to the designers, it's making the disk appear locally > attached so it has to live with the drive timeout limits of a locally > attached hard drive. It's not just LPX, have you tried running iSCSI > over a wireless network :-) > > > > -- > Nik Simpson
First
|
Prev
|
Next
|
Last
Pages: 1 2 3 4 Prev: Help rebuild SATA Mirrored Pair Next: The Windows 2000/XP USB 128 GB problem |