TAO/docs/releasenotes/amh.html

   1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
   2 <html>
   3 <head>
   4    <title>AMH Status</title>
   5     <!--  -->
   6 </head>
   7 <body bgcolor="#FFFFFF" text="#000000">
   8
   9 <h3>TAO's Asynchronous Method Handling (AMH) Feature</h3>
  10     Point of contact: <a href="mailto:coryan@uci.edu">Carlos O'Ryan, </a>
  11                       <a href="mailto:mayur@ics.uci.edu">Mayur Deshpande</a>
  12
  13     <p> The purpose of this document is to provide a guide to using the AMH
  14 capability available in TAO.  This capability is not specified by any
  15 CORBA specification (as of date). This document lists the motivation for AMH,
  16 how to use it and its current implementation status. </p>
  17
  18 <HR SIZE=4 WIDTH="100%">
  19
  20
  21 <h3>Context</h3>
  22 For many types of distributed systems, the CORBA asynchronous method
  23 invocation
  24 <a href="index.html/#ami"> (AMI) </a>
  25 mechanism can improve concurrency, scalability, and
  26 responsiveness significantly. AMI allows clients to invoke multiple
  27 two-way requests without waiting for responses. The time normally
  28 spent waiting for replies can therefore be used to perform other
  29 useful work. The TAO asynchronous method handling (AMH) is a
  30 mechanism, which extends the concepts of AMI from clients to
  31 servers. Servers with AMH capability can return immediately
  32 from (potentially) long, blocking requests. This makes the
  33 servers capable of higher throughput. For a detailed description
  34 for the motivation and potential use-cases of AMH, please refer to
  35 <a href="http://www.dre.vanderbilt.edu/~schmidt/PDF/AMH.pdf">AMH-Motivation. </a>
  36
  37 <HR SIZE=4 WIDTH="100%">
  38
  39
  40 <h3>Using AMH</h3>
  41 Servers that want to take advantage of AMH capability must use
  42 AMH-servants derived from AMH-skeletons. Invoking the
  43 <a href="index.html#idl">IDL-Compiler </a>
  44 with the "-GH" option causes the IDL-Compiler to generate
  45 AMH-skeletons for all interfaces in the ".idl" file.  The signatures
  46 of the AMH-methods differ from the original interfaces as per the
  47 rules described in
  48 <a href="http://www.dre.vanderbilt.edu/~schmidt/PDF/DOA-02.pdf"">AMH-In-Detail. </a>
  49 Servants derived from the AMH-skeletons are registered with the POA in the usual manner
  50 and appear to clients as "normal" servants for the corresponding
  51 interfaces; Thus even while being transparent to the client, these
  52 servants are now capable of handling requests asynchronously.
  53 AMH-servants can be registered in any POA and can intermix with
  54 "normal" servants in the same POA. The granularity of AMH is at
  55 the interface level and not at the method level; Thus for an interface
  56 compiled with the "-GH" option, all operations in that interface are
  57 asynchronous. Currently, AMH doesn't work in conjunction with
  58 <a href="index.html#interceptor">Interceptors </a>
  59 and behaviour is undefined if both are used together.
  60
  61 <HR SIZE=4 WIDTH="100%">
  62
  63
  64 <h3>Current Status</h3>
  65 <b>(As of March 3rd, 2002)</b>
  66 <br>
  67 <P>
  68 <b>Completed:</b>
  69 <ul>
  70 <li>Changes to ORB for asynchronous upcalls.</li>
  71 <li>Generation of AMH-skeletons.</li>
  72 <li>Generation of AMH ResponseHandlers.</li>
  73 <li>AMH Exceptions.</li>
  74 <li>Fix the narrow method bug for AMH skeletons : Fixed
  75 <li>Sending intelligent exception codes back to client in case
  76 something nasty happens in the AMH server.</li>
  77 <li>Example of using AMH available at ACE_wrappers/TAO/examples/AMH/Sink_Server.</li>
  78 <li>Test for AMH : ACE_wrappers/TAO/performance-tests/Latency/AMH</li>
  79 <li>Test for AMH Exceptions : ACE_wrappers/TAO/tests/AMH_Exceptions</li>
  80 </ul>
  81 </P>
  82
  83 <HR SIZE=4 WIDTH="100%">
  84
  85
  86 <h3>Credits</h3>
  87 Original idea first proposed by
  88 <a href="mailto:coryan@uci.edu">Carlos O'Ryan </a>
  89 and changes to ORB for AMH first attempted by
  90 <a href="mailto:brucetrask@aol.com">Bruce Trask. </a>
  91 First go to get the IDL-Compiler to generate AMH code attempted by
  92 <a href="mailto:brunsch@doc.ece.uci.edu"> Darrell Brunsch. </a>
  93 ORB changes for AMH as they stand today put in by
  94 <a href="mailto:mayur@ics.uci.edu">Mayur Deshpande. </a>
  95 Second go at getting IDL-compiler to generate code attempted by Mayur and
  96 <a href="mailto:parsons@cs.wustl.edu">Jeff Parsons. </a>
  97 Getting the IDL-Compiler beast to generate the right AMH code finally
  98 achieved by Carlos.
  99
 100 <HR SIZE=4 WIDTH="100%">
 101
 102
 103 </body>
 104 </html>