hama-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Andronidis Anastasios <andronat_...@hotmail.com>
Subject Re: Lambda architecture
Date Sun, 13 Apr 2014 20:29:45 GMT
I tend to I agree with edward.

Though, I want to say that my envision for hama would include an ad-hoc way to run code on
bsp.

I can explain my self only on the Graph API, as I don't really have a complete idea...

What I was thinking is, that it would be great when a user starts processing a graph and Hama
has load all the vertices in-memory; the user could be able to change the code that is running
on the job. So the values of each vertex, the objects of the vertices, etc... stay in memory
and are ready for process, without losing their state.

Anastasis

On 13 Απρ 2014, at 12:47 μ.μ., Edward J. Yoon <edwardyoon@apache.org> wrote:

> Sure.
> 
> The ignorant people tends to focus only on large-scale data
> processing. So, the example of Lambda architecture consists of HDFS,
> MapReduce batch, HBase, Impala, Storm. However, I think the key of big
> data analytics is in information-flow networks, ..., etc. The HDFS,
> Spark or BSP, ... , and HAMA-883. I'm sure this is more ideal than old
> one. :-)
> 
> 
> 
> On Sat, Apr 12, 2014 at 4:29 AM, Tommaso Teofili
> <tommaso.teofili@gmail.com> wrote:
>> Hi Anastasis,
>> 
>> that's an interesting effort, how do you envision Hama in such an
>> architecture model?
>> 
>> Regards,
>> Tommaso
>> 
>> 
>> 2014-04-11 20:34 GMT+02:00 Andronidis Anastasios <andronat_asf@hotmail.com>:
>> 
>>> FYI:
>>> 
>>> http://lambda-architecture.net
>>> 
>>> Are we interesting in promoting Hama as a part of such an architecture?
>>> 
>>> Cheers,
>>> Anastasis
>>> 
> 
> 
> 
> -- 
> Edward J. Yoon (@eddieyoon)
> Chief Executive Officer
> DataSayer Co., Ltd.


Mime
View raw message