airavata-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Marlon Pierce <>
Subject Re: Need Help !!
Date Mon, 26 May 2014 17:26:51 GMT
Hi Gagan--

I think you are looking for GFAC, which provides the connection between
Airavata and the backend resource.  There is a description of GFAC here
[1].  In short, GFAC has two cooperating plugins: a provider is a client
to a remote resource (Grid, cloud, etc) and a handler supports the
provider to do specialized tasks for a specific scenario. 

We have a simple tutorial document at [2] on writing handlers (this
should move to the wiki soon), and you can find several example
providers in the GFAC module of Airavata.




On 5/26/14 1:15 PM, Gagan Juneja wrote:
> Hi Marlon,
> This make sense. I will take some time to understand PHP code but meanwhile
> what I understood is We have two apis.
> Client api helps us in creating Experiment and other stuff and put them
> into some database.
> Orchestrator api helps in submitting these experiments as jobs for
> execution.
> Obviously we are going to perform some business logic or some computational
> stuff on input data that we set as part of experiment. But where are we
> writing that code?
> As far as Cloud is concern do let me know when you think of running this
> over Amazon Cloud. I am working with Amazon cloud for last 6 months and
> have good experience on automating stuff as well like one click EC2 cluster
> spawning etc.
> Regards,
> Gagan
> On Mon, May 26, 2014 at 9:27 PM, Marlon Pierce <> wrote:
>> Hi Gagan--
>> Thanks for the questions and please keep them coming.
>> Airavata 0.11 from late last year is our latest release, and the trunk
>> is now very different. Airavata 0.11 is the last version with support
>> for XBaya and an older version of the client API (developed mostly by
>> Saminda) that has full support for workflows.
>> We made decisions after Airavata 0.11 to make some major changes.
>> * Concentrate on Science Gateway use cases at the expense of workflows
>> in the near term. Gateways typically run single jobs through a web
>> browser. Scaling and multi-tenancy are the challenges.  Hopefully "near
>> term" is drawing to a close.
>> * Introduce a new component called the Orchestrator that manages job
>> submissions and provide a simpler mechanism for single job submissions
>> (see previous bullet).  Previously, a user would need to define an
>> entire workflow for a single job, which was pretty complicated for most
>> of our driving use cases.
>> * Define our API in Thrift. This has several advantages in addition to
>> the multi-language support.  Thrift can support richer data models, API
>> methods, and exceptions than REST, which was a good fit for our project.
>> * Make significant changes to the Registry to support the new API.
>> * Put workflow support on lower priority until the Thrift API, Registry,
>> and Orchestrator were all stable.
>> These have resulted in several changes that are finally getting wrapped
>> up and will be released in Airavata 0.12 (tentatively in mid-June).
>> After this, we will get back to more frequent releases and also
>> resurrect the workflow work.  Besides workflows, revising the way
>> Airavata manages its application and resource descriptions (called the
>> Application Catalog) will be a big focus.
>> Airavata typically gets used to submit jobs to Grids like the NSF's
>> XSEDE, which use Globus GRAM, UNICORE, or GSI-SSH clients.  Other
>> resources (clouds, the Open Science Grid, non-US Grids) are interesting
>> to us but we just don't have the resources to look at these much. These
>> are interesting places for contributions.
>> At this moment, the most actively developed Thrift clients are
>> temporarily outside Airavata's Git repo, although these should be merged
>> back soon:
>> *
>> *
>> Marlon
>> On 5/26/14 11:30 AM, Gagan Juneja wrote:
>>> Thanks Saminda and Marlon for your prompt response on my query.
>>> I understand from Airavata is "Airavata is a platform which helps user to
>>> execute his workflow over any distributed environment it could be Hadoop
>> or
>>> anything else of same sort." Is this understanding correct.
>>> I am having following question.
>>> 1. How this project is being used by the users or client any brief idea.
>>> Are we exposing cloud, grids as a service or user needs to implement on
>>> their on.
>>> 2. I have looked at class Here most of
>> the
>>> work is like creating an experiment and launching it.  I am not able to
>>> find any thing related to workflow or job that we want to run this input
>>> data that we are setting in Experiment object. Where can we create Jobs,
>>> tasks or workflows?
>>> 3. Why do Xbaya is deprecated and replaced with thrift API? (Anything
>>> special apart from multi language support.)
>>> I have lot more question will keep bothering you :).
>>> Regards,
>>> Gagan
>>> On Sun, May 25, 2014 at 11:31 PM, Marlon Pierce <> wrote:
>>>> The QuickStart info is out of date, and we will be updating it soon.
>>>> After compiling and starting the server, you can try out PHP sample
>>>> scripts in
>> ./airavata-api/airavata-client-sdks/airavata-php-sdk/src/main/resources/samples/
>>>> Marlon
>>>> On 5/24/14 4:13 AM, Saminda Wijeratne wrote:
>>>>> XBaya is deprecated in the upcoming release. Please find all the
>> release
>>>>> artifacts under modules/distribution/release/target/release-artifacts.
>>>>> On Fri, May 23, 2014 at 11:14 PM, Gagan Juneja <
>>>>>> Hi Team,
>>>>>> I am new to Airavata project. I was following Quick Start document.
>>>>>> build Airavata project using default profile. I did not find any
>>>> artifact
>>>>>> in modules/distribution target directory and even many of the modules
>>>> did
>>>>>> not build such as xbaya.
>>>>>> Please guide me how to start with this.
>>>>>> Thanks & Regards,
>>>>>> Gagan

View raw message