> I think having a Pregel-like API would be good but I'd like more to add such > an API rather than replacing the current BSP-oriented one.
+1 On Wed, Jun 29, 2011 at 5:33 PM, Tommaso Teofili <[email protected]> wrote: > I think having a Pregel-like API would be good but I'd like more to add such > an API rather than replacing the current BSP-oriented one. > > 2011/6/29 Eric Charles <[email protected]> > >> Yes, from Hadoop Summit Page, I read: "Similar to Pregel, Giraph follows >> the bulk-synchronous parallel model relative to graphs where vertices can >> send messages to other vertices during a given superstep." >> >> If you don't explicitly mention Pregel in Hama headline, there are less >> change you will be heard. >> > > I agree, it may be worth to have something about it on the wiki. > Regards, > Tommaso > > >> >> Thx >> >> >> On 29/06/11 08:12, Thomas Jungblut wrote: >> >>> Hey, >>> >>> just have seen JPregel. Don't think they going to get far, since this is >>> just a 1:1 java-portation of the pregel paper. >>> Giraphs presentation is today, are there any live feeds of the hadoop >>> summit?:) >>> >>> I'm +1 for pregel-like API because the "thinking like a vertex" is always >>> more intuitive than BSP itself. If we can manage to add, let's call it TLV >>> (Thinking Like a Vertex) AND to keep the BSP engine as itself, Hama would >>> be >>> very powerful. >>> >>> Greetings >>> >>> >>> 2011/6/29 Edward J. Yoon<[email protected]> >>> >>> Hi, >>>> >>>> Recently, I'm see many Pregel clone projects e.g., giraph, goldenorb, >>>> jpregel, .., etc. (although they are all the early stage) >>>> >>>> I still believe that the core is a BSP computing engine but, "Pregel >>>> clone" is a good marketing strategy. >>>> >>>> Should we plan the Pregel-like graph APIs for 0.4-incubating? >>>> >>>> Thanks. >>>> -- >>>> Best Regards, Edward J. Yoon >>>> @eddieyoon >>>> >>>> >>> >>> >>> >> -- >> Eric >> > -- Best Regards, Edward J. Yoon @eddieyoon
