Update docs/platform_basics_brainstorming_2019-04-17.md

docs/platform_basics_brainstorming_2019-04-17.md

@@ -1,39 +1,65 @@
-## Manager Service:
-- Joblist
-    - Thread safe
-    - REST API (https://github.com/zalando/connexion)
-        - CreateNewJob
-        - DelteJobs
-        - GetJobInfo
-        - AlterJob
-    - Scheduling
-    - Ressource management
-- manages also files on file server 
-
 ## Web Server
-- serves content.
+- **Solutions**: 
+    - Apache (with mod WSGI)
+    - or nginx (with gunicorn I guess)
+- Serves content
+- handels HTTP requests etc.
 - serves forms for user request and inputs
 - has copy of Joblist to display those for the user according to requests etc.
 - talks to the Manager service
 - Users CANNOT talk directly with the manager
 - has list of all currently running user sessions (maybe used for authentication)
 
-## Authentication via LDAP (not sure if we need that)
+## Application Server
-- university internal authentication with LDAP
+- **Solution**: Flask
-- LDAP and connexion can use (OAuth 2)
+
-    - https://ldapwiki.com/wiki/Best%20Practices%20for%20LDAP%20Security
+### Authentication and session management
-    - https://connexion.readthedocs.io/en/latest/security.html?highlight=authentication
+ - **Solutions**:
-- university external authentication with something else
+    - Flask-Login (minimal)
+    - Flask-Session (maybe a bit more functions)
+- handels both internal and external users
+- Relational Database
+    - **Solutions**_
+        - PostgreSQL
+        - MariaDB
+- Object Relational Mapper
+    - **Solutions**:
+        - Flask-SQLAlchemy
+
+### Manager Service:
+- Part of the Application Server
+- Joblist
+    - **Solution**:
+        - http://www.celeryproject.org/
+        - Thread safe
+        - Scheduling
+        - Ressource management
+- REST API
+    - **Solution**
+        - Flask internal
+    - Passes requests to the joblist/celery
+    - Functions:
+        - create_job
+        - delete_job
+        - get_job (JSON Object or metadata or both?)
+        - alter_job
+- manages also files on file server 
+
 
 
 ## Compute pool: Docker Cluster
+- Solutions: 
+    - Kubernetes
+    - swarm 
 - How to handle job and resource management for user Jobs and processes.
 - gets requests and tasks from the manager
 
-## File Server
+## File Server (Scans, pdfs etc.)
 - stores user input and
 - output files
 - Maybe WebDAV
 
 ## Database
 - document based data base like mongoDB preferred
+- could be possible that a python framework for jobs and tasks exist
+mongo Db would not be necessary then