+1 On Thu, Jul 7, 2011 at 11:52 AM, ChiaHung Lin <[email protected]> wrote: > I agree, we first work on checkpoint as it is a fundamental feature. Later on > the support for HNG can be added. > > -----Original message----- > From:Edward J. Yoon <[email protected]> > To:[email protected],[email protected] > Date:Wed, 6 Jul 2011 20:30:35 +0900 > Subject:Re: [DISCUSS] "Fault Tolerant" system design and consideration of > compatibility for the hadoop nextgen or mesos platform. > > I guess, HNG won't stabilized until late 2011. We don't need to wait, > or depend on HNG. > > But, I think, the Hama job can be submitted to both Hama cluster and HNG. > > We don't need to choose one path. > > On Wed, Jul 6, 2011 at 11:45 AM, ChiaHung Lin <[email protected]> wrote: >> I was under the impression that the previous discussion decided Hama would >> have its own fault tolerance mechanism, including master and workers; >> therefore, HNG is not taken into account. >> >> Regardless of using HNG or creating Hama's style mechanism for fault >> tolerance, checkpoint is required so we will need to implement this feature >> first before processing the next step. >> >> -----Original message----- >> From:Edward J. Yoon <[email protected]> >> To:[email protected] >> Date:Fri, 1 Jul 2011 16:31:34 +0900 >> Subject:Re: [DISCUSS] "Fault Tolerant" system design and consideration of >> compatibility for the hadoop nextgen or mesos platform. >> >> I wonder whether Hama job can't be executed with MR2 if we have own >> BSP computing engine. >> >> If so, what's the reason we can't support both MR2 version and own BSP >> cluster version? >> >> >> >> -- >> Best Regards, Edward J. Yoon >> @eddieyoon >> >> >> >> -- >> ChiaHung Lin >> Department of Information Management >> National University of Kaohsiung >> Taiwan >> > > > > -- > Best Regards, Edward J. Yoon > @eddieyoon > > > -- > ChiaHung Lin > Department of Information Management > National University of Kaohsiung > Taiwan >
-- Best Regards, Edward J. Yoon @eddieyoon
