HA mode bonding for IB over SDP

HA mode bonding for IB over SDP

[Bhavin]

Hi,

I am a student pursuing masters. I have chosen to work on infiniband for my
final year project. As part of my project I am trying to implement infiniband
over SDP. I have also setup two systems for this purpose having Mallanox IB
cards. I found that OFED doesn’t have support for HA mode failover for IB over
SDP when searching for limitations of this configuration. So I have decided to
work on this issue.

Right now I am going through the OFED source code to understand how different
functionalities are implemented. I also found there is something called
APM(Automatic Path Migration) is present in OFED.

Is it possible to use APM for failover with IB over SDP? If yes, please guide me
on how can I use APM for IB over SDP.

Please Help.

Thanks


--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: HA mode bonding for IB over SDP

[Moni Shoua]

If you want a reference code that works you can browse cmpost (get
from git clone git://git.openfabrics.org/~shefty/rdma-dev.git -
chechout branch test-apps).

There are several design issues that you need to resolve before you
start to implement.
Examples:
1. Where to add the code to
   - SDP: via direct calls to APM functions in CM
   - RDMA_CM: enhance interface
2. How to determine endpoints of alt path
3. What to do when migration occurs
4. exchange alt path info during connection establishment or after
connection establishment (LAP)

If you have specific questions then we'd be happy to share our
experience and try to answer.


If you want a reference code that works you can browse cmpost (get
from git clone git://git.openfabrics.org/~shefty/rdma-dev.git -
chechout branch test-apps).

There are several design issues that you need to resolve before you
start to implement.
Examples:
1. Where to add the code to
   - SDP: via direct calls to APM functions in CM
   - RDMA_CM: enhance interface
2. How to determine endpoints of alt path
3. What to do when migration occurs
4. exchange alt path info during connection establishment or after
connection establishment (LAP)

If you have specific questions then we'd be happy to share our
experience and try to answer.



On Fri, Apr 1, 2011 at 8:11 AM, Bhavin <bhawinkumar.thumar-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
> Hi,
>
> I am a student pursuing masters. I have chosen to work on infiniband for my
> final year project. As part of my project I am trying to implement infiniband
> over SDP. I have also setup two systems for this purpose having Mallanox IB
> cards. I found that OFED doesn’t have support for HA mode failover for IB over
> SDP when searching for limitations of this configuration. So I have decided to
> work on this issue.
>
> Right now I am going through the OFED source code to understand how different
> functionalities are implemented. I also found there is something called
> APM(Automatic Path Migration) is present in OFED.
>
> Is it possible to use APM for failover with IB over SDP? If yes, please guide me
> on how can I use APM for IB over SDP.
>
> Please Help.
>
> Thanks
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: HA mode bonding for IB over SDP

[Moni Shoua]

Thanks a lot for providing me such useful details.

Besides the APM functionality one thing I don't understand is that why is the IB-bonding not working with SDP, as SDP also uses IPoIB for addressing. Can you please describe about the problem which doesn't allow the fail-over to take place with IB-bonding driver? This would be more helpful in tackling the problem.

Again, I appreciate for your valuable information.

Thanking You,
Bhavin  

Hi Bhavin  
Bonding enslaves netdev devices (i.e. IPoIB interfaces) but SDP doesn't send/recv traffic on these interfaces.
SDP uses to IB interfaces to send/recv traffic to the network and socket interface to send/recv data from/to application.
However, IPoIb interfaces are used for address resolution so you can't give them up completely.

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html