2001 Sun Microsystems

Transcription

1 The Project A Future Vision for the Web, and Some Tools to Get There Stephen Uhler Sun Microsystems Laboratories Sun Microsystems

2 Overview A vision for the future of the Web An architecture for the vision Brazil - a sample implementation Sample applications Using the toolkit Summary 2 Overview

3 Research is like driving a car at night. You can only see as far as your headlights, but you can make the whole trip that way. - E.L. Doctorow 3 Overview

4 The Web Today Same client for every application ¾ browser Content supplied by the server: ¾ Web server ¾ Business logic ¾ Database/transactions 4 Overview

5 A Vision Distribute the application throughout the network 5 Vision

6 A Vision Distribute the application throughout the network Support different clients pager pda 6 Vision

7 A Vision Distribute the application throughout the network Support different clients Provide content aggregation for small data sources pager pda 7 Vision

8 A Vision pager pda Distribute the application throughout the network Support different clients Provide content aggregation for micro servers Extend reach of applications to external content 8 Vision

9 Some Examples Accounting Catalog XML HTML Workflow ¾ Legacy application provides a generic catalog interface ¾ Portal proxies and filters content, mediating with the accounting system ¾ Company discounts, billing info added 9 Vision

10 Some Examples Workflow HTML HTML/XML storage Localization ¾ Legacy application provides a generic user interface ¾ Portal proxies and filters content, adding localization 10 Vision

11 Some Examples Workflow Localization Persistent object store Personalization ¾ Content filtered by user preferences stored on portal. ¾ Transcoding for alternate GUI's Desktop browser pager pda 11 Vision

12 Some Examples Desktop browser Request Response Workflow Localization Personalization Device Integration ¾ ¾ Request filtered by portal ¾ Content gathered from multiple sources ¾ Integrated content delivered to client 12 Vision

13 Breadth of Computing Environments Servers $10,000+ Ram 1GB Storage 1TB Power wall IP 3 Java 3 GUI 2 pager pda 13 Vision

14 Breadth of Computing Environments Servers $10,000+ Desktops $1,000 Ram 64MB Storage 10GB Power wall IP 3 Java 3 GUI 3 pager pda 14 Vision

15 Breadth of Computing Environments Servers $10,000+ Desktops $1,000 PDA s $100 Ram 2MB Storage 2MB Power Battery IP 3 Java 3 GUI 3 pager pda 15 Vision

16 Breadth of Computing Environments Servers $10,000+ Desktops $1,000 PDA s $100 Embedded $10 Ram 0.5MB Storage 0.5MB Power wall IP 3 Java 3 GUI 2 pager pda 16 Vision

17 Breadth of Computing Environments pager pda Servers $10,000+ Desktops $1,000 PDA s $100 Embedded $10 Micro $1 Ram 0.05KB Storage 2KB Power battery IP 2 Java 2 GUI 2 17 Vision

18 A common infrastructure for a wide range of systems: ¾ Small applications are simple to do ¾ Large applications achieved by combining simple parts in consistent ways A methodology that encourages reuse at the appropriate granularity to manage complexity. Rich toolkit of powerful, composable parts. 18 Architecture

19 Brazil Traditional content services pager pda 19 Architecture

20 Brazil Traditional content services Dynamic content transcoding for different GUI s pager pda 20 Architecture

21 Brazil Traditional content services Dynamic content transcoding for different GUI s Content aggregation for simple data sources pager pda 21 Architecture

22 Brazil Traditional content services Dynamic content transcoding for different GUI s Content aggregation for simple data sources Extraction and integration of foreign content pager pda 22 Architecture

23 Brazil pager pda Traditional content services Dynamic content transcoding for different GUI s Content aggregation for simple data sources Extraction and integration of foreign content Communication with Sub- IP Devices via proxies 23 Architecture

24 Architecture Comparison: How Complexity Increases The "Other" Brand API time Components 24 Architecture

25 Brazil Functional Components Content retrieval Request infrastructure Content transformation Special Purpose Support 25 Implementation

26 Content retrieval Files/Templates CGI Devices Virtual Web Sites 26 Implementation

27 Request Infrastructure Session management Authentication Access control Request distribution 27 Implementation

28 Content Transformation Content filtering XML template processing Data extraction and manipulation TCL, Python, SQL, LDAP, IMAP Content formatting: BSL (Brazil Scripting Language) 28 Implementation

29 Special Purpose X10 Jini SmartCards Certificate Authorities ActiveX/Com connectivity JRMS Gnutella 29 Implementation

30 Support XML processing URL rewriting Templates HTTP/HTML utilities Regular expressions Proxy services 30 Implementation

31 Sample Portal Configuration (conceptual view) Pages are retrieved from remote sites. Content is extracted A new page is created, combining extracted content with desired look. Content Server Content Extraction Filters Presentation Templates To client 31 Sample Applications

32 32 Portal Configuration (concrete view) Session Manager Server Preferences Not Found Local content Legacy CGI Virtual sites Access control Form filler pjama Content retrieval M O N P Content delivery Plain text to html Table of Contents Navigation Bar Content extraction Template selection Content Integration Sample Applications Content filtering

33 FridgeNet (conceptual view) Pickle Net Powerline Residential Gateway Personal Appliance Portal 33 Sample Applications

34 FridgeNet in Operation 34 Sample Applications

35 35 Sample Applications

36 home Residential Gateway Pickle Net Interface 36 Sample Applications

37 How to get Started Download your own copy ¾ View the built-in documentation ¾ run: java -jar brazil.jar ¾ Point your browser to: Add brazil.jar to your CLASSPATH Unpack the archive ¾ Sample applications ¾ Source code 37 Using Brazil

38 How to get Started New combinations of existing pieces Write your own application specific code Provided by: ¾Handlers*... Server ¾Filters.. FilterHandler ¾Templates*. TemplateFilter ¾Scripting Tcl -or- Python Template 38 Using Brazil

39 Existing Application Class MyApp { public static void main(string[] args) { } } 39 Using Brazil

40 Hello World (handler version) Class MyApp implements Handler { public boolean init(server server, String name) { return true; } public boolean respond(request request) { request.sendresponse("hello World!"); return true; } public static void main(string[] args) { new Server(new ServerSocket(8080), MyApp, new Properties()).start(); }} 40 Using Brazil

41 Hello world (template version) Source code Class MyApp extends Template { public void tag_myapp(rewritecontext rc) { rc.append("hello World!"); }} Sample template <title>world Existence Page</title> <body> <h1><myapp></h1> </body> 41 Using Brazil

42 Using Templates (source code) Application specific functionality ¾ Invent application specific query language ¾ Return results as name/value pairs ¾ No presentation markup Class MyApp extends Template { public void tag_myapp(rewritecontext rc) { String p=rc.get("prefix"); rc.request.props.put(p + ".h", "Hello"); rc.request.props.put(p + ".w", "World"); }} 42 Using Brazil

43 Using Templates (template) Presentation ¾ Call application specific query language ¾ Use templates to generate presentation <title>world Existence Page</title> <myapp prefix=world> <ul> <foreach name=i glob=world.*> <li><property name=i> </foreach> </ul> 43 Using Brazil

44 Using Templates (configuration) Integration ¾ Use server properties to tie functionality together templates= MyApp \ sunlabs.brazil.template.propstemplate \ sunlabs.brazil.template.bsltemplate public static void main(string[] args) { Properties p = ; new Server(new ServerSocket(8080), TemplateHandler, p).start(); }} 44 Using Brazil

45 Summary Advantages ¾ Composability: Parts do fit together ¾ Reusability: Special purpose handlers reused ¾ Slicers and Formatters allow separation of content from presentation ¾ Less structure: Easy to get started Disadvantages ¾ Performance: Everything funnels through small interface ¾ Security: Uniform access to all information ¾ Less structure: Less framework for large systems 45 Summary

46 Some Numbers Size ¾ 25k lines of code 1 in core system ¾ 75k lines of code 1 in sample applications ¾ 30k 250k bytes Jar file size Performance 2 ¾ 700 hits/sec (500 if every page is rewritten) ¾ 14Mbytes/sec throughput Distribution ¾ 10,000+ copies downloaded 1 Java Programming Language (JDK1.1) 2 Sun 420R (entry level server) 46 Summary

47 References (internal to Sun) javaworld/jw /jw-0811-javadev.html Netscape Communications Corporation Logo is a registered trademark in the United States and other countries UltraSPARC is a registered trademark of SPARC International, Inc. 47 Summary

48 The End Brazil Team: Rinaldo DiGiorgio Colin Stevens Stephen Uhler