|
Home > Archive > Data Storage > August 2005 > CRC of an out-of-order arrival data.
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
CRC of an out-of-order arrival data.
|
|
|
| Hi;
I figured out a method to calculate accumulated CRC of data when its
packet arrival is out of order without large intermediate data buffer.
In aother word, I can calculate each packet's contribution to the final
CRC using each packet's CRC which could be give by hardware or protocal
chip. The overhead is little and the complexity is O(N), where N is the
packet number. By this method, what ever implemented in software or
hardware, it can potentially save lot of buffer space, CRC engine and
huge memory bandwidth. I thought iSCSI storage system could be a ideal
application candidate. I do not know if this one has some real value or
be implemented alreday. Any comment or feedback is welcome!
| |
| Maxim S. Shatskih 2005-08-16, 2:46 am |
| Out of order arrival is usually a sign of a packet being lost, which is -
in TCP - usually due to congestion on the router. Such things are very rare on
iSCSI.
--
Maxim Shatskih, Windows DDK MVP
StorageCraft Corporation
maxim@storagecraft.com
http://www.storagecraft.com
"alpha" <zhg.liu@gmail.com> wrote in message
news:1124150106.352296.268030@g44g2000cwa.googlegroups.com...
> Hi;
>
> I figured out a method to calculate accumulated CRC of data when its
> packet arrival is out of order without large intermediate data buffer.
> In aother word, I can calculate each packet's contribution to the final
> CRC using each packet's CRC which could be give by hardware or protocal
> chip. The overhead is little and the complexity is O(N), where N is the
> packet number. By this method, what ever implemented in software or
> hardware, it can potentially save lot of buffer space, CRC engine and
> huge memory bandwidth. I thought iSCSI storage system could be a ideal
> application candidate. I do not know if this one has some real value or
> be implemented alreday. Any comment or feedback is welcome!
>
| |
| Rob Turk 2005-08-16, 2:46 am |
| "alpha" <zhg.liu@gmail.com> wrote in message
news:1124150106.352296.268030@g44g2000cwa.googlegroups.com...
> Hi;
>
> I figured out a method to calculate accumulated CRC of data when its
> packet arrival is out of order without large intermediate data buffer.
> In aother word, I can calculate each packet's contribution to the final
> CRC using each packet's CRC which could be give by hardware or protocal
> chip. The overhead is little and the complexity is O(N), where N is the
> packet number. By this method, what ever implemented in software or
> hardware, it can potentially save lot of buffer space, CRC engine and
> huge memory bandwidth. I thought iSCSI storage system could be a ideal
> application candidate. I do not know if this one has some real value or
> be implemented alreday. Any comment or feedback is welcome!
>
If you can demonstrate the cases where this is useful, write an
implementation and submit it to the iSCSI Linux developers. I'm sure that if
you have strong arguments, they will add it to their code for inclusion.
Rob
| |
|
| The issue is system need buffer arrival data until they were all
received since they could come out-of-order, even it is rare as you
mention. Then system need read them out to verify CRC, which took
memory bandwith too.
It will be great if storage system only save each packet's contribution
to the final CRC and commit this packet's data immediately. Util all
packets were received, verify over all CRC. No intermediate buffer for
data in this approcah. This is the point.
Maxim S. Shatskih wrote:[vbcol=seagreen]
> Out of order arrival is usually a sign of a packet being lost, which is -
> in TCP - usually due to congestion on the router. Such things are very rare on
> iSCSI.
>
> --
> Maxim Shatskih, Windows DDK MVP
> StorageCraft Corporation
> maxim@storagecraft.com
> http://www.storagecraft.com
>
> "alpha" <zhg.liu@gmail.com> wrote in message
> news:1124150106.352296.268030@g44g2000cwa.googlegroups.com...
|
|
|
|
|