Hi All, As Chris mentioned, I am a MSc student planning to work on Apache Airavata and Apache OODT integration as part of my MSc Project. Up to now, I have identified the following integration points to leverage benefits from Apache OODT into Apache Airavata.
1. Apache Airavata GFac component has been designed and implemented with several extension points. There are Pre and Post execution chains that executes prior and after the service execution by computational resource provider. CAS PGE Task Wrapper performs file staging, configuration file creation prior to execution of task. It invokes Crawler to ingest output into product files together with Metadata. Tasks that PGE task wrapper performs prior to execution of a task can be implemented as a task in pre chain in GFac. Tasks that are executed after the workflow task execution can be implemented as post execution chain in GFac. 2. Apache Airavata provides provenance aware workflow where workflow processing can skip processing some tasks if the same task has been already processed and output is available. Rich Metadata catalog plays a key role in provenance aware workflow execution. Apache OODT could help Airavata to improve the current implementation of provenance aware workflow processing. 3. Apache OODT product and profile servers provides robust mechanism for data search and discovery from a complex heterogeneous data sources. Integration of Apache OODT product and profile servers with Apache Airavata would help to discover required resources prior to execution of a task. It may be possible to introduce new Input components to take XML query as Input to discover relevant data using configured profile and product servers and execute tasks with data matched with input query. Cheers, Sanjaya Hi Ray, and Suresh, I'm happy to discuss Airavata and OODT integration paths too — what did you have in mind? There are a few research efforts going on mainly at the level of some MS student projects, but perhaps you can help focus the integration. Thanks! Cheers, Chris From: Suresh Marru <[email protected]<mailto:[email protected]>> Reply-To: "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>> Date: Monday, January 21, 2013 6:11 PM To: "[email protected]<mailto:use [email protected]>" <[email protected]<mailto:[email protected]>> Subject: Re: what should be in airavata? Hi Ray, Its not our goal to make the reader work for what they want to know, but this ambiguity is certainly evident on a evolving open source project like Airavata so contributors like yourself will find natural first place to jump on board and fix the things you do not like :) So thank you for your contribution to Airavata (very seriously), this is great feedback, we will even appreciate if you can further contribute your notes or fix documentation as you work through so future users will benefit. Referring to Airavata high level architecture [1] Airavata has three levels of components, the server, a built in GUI and an API (airavata client). We do not have yet built any web interfaces which is where we would like to see Airavata integrated with Rave. We very much would like to see Airavata integrated with OODT and there have been some discussions but no concrete tasks have been accomplished. It will be great if you can detail what you would like to do a little more. Your README feedback is well taken, we will take into account for the upcoming release 0.6 which is going through testing right now (in dev mailing list). Meanwhile you may want to refer to wiki to see if the documentation you are looking for exists there. Here are some OODT integration ideas [3] [4] Suresh [1] - http://airavata.apache.org/architecture/overview.html [2] - https://cwiki.apache.org/confluence/display/AIRAVATA/index [3] - https://issues.apache.org/jira/browse/AIRAVATA-342 [4] - http://markmail.org/message/5uabiaceuj2eqayl On Jan 21, 2013, at 6:55 PM, Ray Martin <[email protected]<mailto:[email protected]>> wrote: version 0.5 download is 52.1Mbytes version 0.4-incubating is 74.6Mbytes version 0.4 has folders including gui and lib version 0.5 does not have the folders gui and lib, but the README says it should. One of the most important parts of documentation is to tell the reader why they may want to do something. The reader has never before seem the functionality that you are offering. Currently apache web sites do not provide sufficient information as to the purpose of what is being provided. For example, I am interested in any work that has been placed into integrating Rave/Wave with Airavata/OODT. Would or should one expect to find that information under airavata-client? I dunno - should I? It is so hard to follow and decipher. Maybe that is the goal - make the reader work for what he wants to know.
