[Saga-devel] saga-projects SVN commit 914: /papers/CPC/

[sjha at cct.lsu.edu]

User: sjha
Date: 2009/01/29 09:28 PM

Added:
 /papers/CPC/
  cpc_notes.txt

Log:
 the outline of the paper (first draft)
   arising from discussions between HK + SJ

File Changes:

Directory: /papers/CPC/
=======================

File [added]: cpc_notes.txt
Delta lines: +63 -0
===================================================================
--- papers/CPC/cpc_notes.txt	2009-01-30 02:36:20 UTC (rev 913)
+++ papers/CPC/cpc_notes.txt	2009-01-30 03:28:35 UTC (rev 914)
@@ -0,0 +1,63 @@
+Scope of the CPC Paper
+
+1.  Background: changing landscape of computational/digital
+  infrastructure:
+  - Traditional  Grids	+ Distributed Systems
+  - Emerging Distributing Systems (Clouds, distributed data-store..)
+  - High-end machines are becoming heterogenous, but at a different
+        level of granularity
+    Implication of the heterogenity of monolithic systems implies
+
+  Develop scientific applications for all of the above -- in a
+  "Standard"  uniform way. This is SAGA i.e, let us not just
+  stay confined to traditional Grids. But we need to be very 
+  clear that SAGA is applicable only for certain classes of 
+  applications; SAGA != MPI; SAGA != a programming model per se
+  and not an execution model (like MPI). SAGA supports
+  different PM and ExecutionModels.
+
+  Motivation: 
+  Traditional Grid Systems:
+    - heterogeneity (of what?)
+    - .. 
+  
+   Emerging Distributed Systems:
+    - heterogeneity (system interface, but also semantic difference)
+
+   High-End Motivation:
+     Why SAGA on these machines?
+
+   SAGA provides the fundamental abstractions --- functionality,... ,
+   that are required to develop applications in a system independent
+   fashion
+ 
+  Portability: syntactic, semantic and platform independence.
+
+  SAGA Programming System:  (Engine + Packages) + Adaptors + Development tools for SAGA
+
+  The API is exposed via the Engine (functional:context, session,
+  non-functional:others) and Packages (functional and non-functional)
+
+
+2. Some Distributed SCIENTIFIC Applications .. 
+   possible use cases..
+
+3. How these features/requirements are supported?
+   Could have overlap with "requirements" 
+   Develop "requirements"
+   ....
+
+4. All the Gory Details that have been developed to support the points
+   in section 3.
+
+   This is where OOPSLA paper comes in handy. Lots of the meat 
+   is the same, but presented in a different perspective and style.
+
+5. Applications that utilize these... and how.. for what.. (Broad spectrum)
+   .. talk about N applications that we've developed
+   .. show the specific features from Section 3 and how they are
+   addressed/utilised/or contribute to the application usage
+  
+What is the contribution of this Paper: OOPSLA Paper is Implementation
+paper + MCS Paper is Interface Paper, this is a mege of the two using
+a consistent vocabulary along with applications.