1. EEC-681/781 Distributed Computing Systems Lecture 5 Wenbing Zhao Department of Electrical and Computer Engineering Cleveland State University wenbing@ieee.org

2. 2 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Outline Remote Procedure Call Remote Method Invocation Case study: Java RMI –Material taken from http://java.sun.com/developer/onlineTraining/rmi/RMI.html

3. 3 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Conventional Procedure Call Parameter passing in a local procedure call: the stack before the call to read() The stack while the called procedure is active count = read(fd, buf, bytes);

4. 4 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Parameter Passing in Procedure Call Passing by value Passing by reference Read() fd = 10; bytes = 1024; buf = 0x01400000; count = read(fd, buf, bytes); buffer fd = 10; bytes = 1024; buf = 0x01400000; 0x01400000

5. 5 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Remote Procedure Call Observations: –Application developers are familiar with simple procedure model –Well-engineered procedures operate in isolation (black box) –There is no fundamental reason not to execute procedures on separate machine Conclusion: communication between caller & callee can be hidden by using procedure-call mechanism

6. 6 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Client and Server Stubs Remote Procedure Call (RPC) achieves distribution transparency by using a client stub and a server stub The client stub provides the interface of the procedure call => illusion of a local call interface –It packs the parameters into a message and requests that message be sent to the server The server stub is a piece of code that transforms requests coming in over the network into local procedure calls

7. 7 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao RPC between Client and Server

8. 8 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Steps of Remote Procedure Call

9. 9 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Parameter Passing for RPC Marshaling: must convert the parameters into a message Marshaling is not trivial: –Client and server machines may have different data representations (think of byte ordering) –Wrapping a parameter means transforming a value into a sequence of bytes –Client and server have to agree on the same encoding –Client and server need to properly interpret messages, transforming them into machine-dependent representations

10. 10 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Passing Value Parameters Original message on the Pentium The message after receipt on the SPARC The message after being inverted. The little numbers in boxes indicate the address of each byte

11. 11 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Parameter Passing A procedure The corresponding message

12. 12 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao External Data Representation Data structures: –“flattened” on transmission –rebuilt upon reception Primitive data types: –Byte order (big-endian: MSB comes first) –ASCII vs UNICODE (2 bytes per character) –Marshalling/unmarshalling To/from agreed external format

13. 13 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao External Data Representation XDR (RFC 1832), CDR (CORBA), Java: –data -> byte stream –object references HTTP/MIME: –data -> ASCII text

14. 14 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao CORBA CDR Example The flattened form represents a Person struct with value: {‘Smith’, ‘London’, 1934} 0–3 4–7 8–11 12–15 16–19 20-23 24–27 5 "Smit" "h___" 6 "Lond" "on__" 1934 index in sequence of bytes4 bytes notes on representation length of string ‘Smith’ length of string ‘London’ unsigned long

15. 15 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Interface Definition Language In order to allow servers to be accessed by differing clients, Interface Definition Language (IDL) is usually used to allow various platforms to call the RPC –The first popular implementation of RPC on Unix was Sun's RPC, which was used as the basis for NFS From IDL, client and server stubs can be generated automatically to facilitate development

16. 16 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Parameter Passing While passing value parameters is relatively straightforward, passing reference parameters is difficult If we introduce a remote reference mechanism, access transparency can be enhanced: –Remote reference offers unified access to remote data –Remote references can be passed as parameter in RPCs

17. 17 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Asynchronous RPC The interconnection between client and server in a traditional RPC The interaction using asynchronous RPC

18. 18 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Asynchronous RPC A client and server interacting through two asynchronous RPCs. This scheme is also called deferred synchronous RPC

19. 19 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Remote Distributed Objects Data and operations encapsulated in an object Operations are implemented as methods, and are accessible through interfaces Object offers only its interface to clients Object server (or host server) is responsible for a collection of objects Server skeleton handles (un)marshaling and object invocation Client stub (proxy) implements interface

20. 20 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Distributed Objects

21. 21 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Client-to-Object Binding Object reference: denotes server, object, and communication protocol. Having an object reference allows a client to bind to an object: –Client loads associated stub code –Stub is instantiated and initialized for specific object Two ways of binding: –Implicit: Invoke methods directly on the referenced object –Explicit: Client must first explicitly bind to object before invoking it

22. 22 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Remote Method Invocation Steps Client invokes method at stub Stub marshals request and sends it to server Server ensures referenced object is active Request is unmarshaled by object’s skeleton, and referenced method is invoked Result is marshaled and passed back to client Client stub unmarshals reply and passes result to client application

23. 23 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao RMI: Parameter Passing Passing objects by value: A client may pass a complete object as parameter value: –An object has to be marshaled: Marshall its state Marshall its methods, or give a reference to where an implementation can be found –Server unmarshals object. Note that we have now created a copy of the original object

24. 24 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao RMI: Parameter Passing Passing objects by reference: Much easier than in the case of RPC: –One can simply bind to referenced object, and invoke methods –Unbind when referenced object is no longer needed

25. 25 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Parameter Passing Passing an object by reference or by value

26. 26 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Java RMI Architecture Design goal: to create a Java distributed object model that integrates naturally into the Java programming language and the local object model The definition of a remote service is coded using a Java interface The implementation of the remote service is coded in a class Interfaces define behavior and classes define implementation

27. 27 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Java RMI Architecture Layers Stub and Skeleton layer: intercepts method calls made by the client to the interface and redirects these calls to a remote RMI service Remote reference layer: to interpret and manage references made from clients to the remote service objects Transport layer: provides basic connectivity, as well as some firewall penetration strategies Transport Layer Remote Reference Layer Stubs & Skeletons Client ProgramServer Program RMI System

28. 28 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Stub and Skeleton Layer Java RMI follows the Proxy pattern –The stub class plays the role of the proxy –The remote service implementation class plays the role of the RealSubject > Subject Request() RealSubject Request() Proxy Request()

29. 29 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Remote Reference Layer Defines and supports the invocation semantics of the RMI connection –This layer provides a RemoteRef object that represents the link to the remote service implementation object –The stub objects use the invoke() method in RemoteRef to forward the method call –The RemoteRef object understands the invocation semantics for remote services

30. 30 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Transport Layer The Transport Layer makes network connection between JVMs Java Remote Method Protocol (JRMP): Java RMI’s wire protocol (proprietary) on top of TCP/IP

31. 31 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao RMI Registry RMI Registry ( rmiregistry ): a simple naming service that comes with Java RMI –It runs on each machine that hosts remote service objects, by default on port 1099 –It accepts registration request only from the local RMI servers –It accepts queries for services anywhere

32. 32 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao RMI Registry To access the remote server from client side: –Query a registry by invoking the lookup() method on the static Naming class –The method lookup() accepts a URL that specifies the server host name and the name of the desired service rmi:// [: ]/ –The method returns a remote reference to the service object

33. 33 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Steps to Create a Remote Service First create a local object that implements that service Next, export that object to Java RMI. When the object is exported, Java RMI creates a listening service that waits for clients to connect and request the service After exporting, register the object in the Java RMI Registry under a public name

34. 34 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Using Java RMI A working Java RMI system is composed of several parts: –Interface definitions for the remote services –Implementations of the remote services –Stub and Skeleton files –A server to host the remote services –An RMI Naming service that allows clients to find the remote services –A class file provider (an HTTP or FTP server) –A client program that needs the remote services

35. 35 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Steps to Build a System Design your system Write and compile Java code for interfaces Write and compile Java code for implementation classes Generate Stub and Skeleton class files from the implementation classes Write Java code for a remote service host program Develop Java code for Java RMI client program Install and run Java RMI system

36. 36 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Service Interface public interface Calculator extends java.rmi.Remote { public long add(long a, long b) throws java.rmi.RemoteException; public long sub(long a, long b) throws java.rmi.RemoteException; public long mul(long a, long b) throws java.rmi.RemoteException; public long div(long a, long b) throws java.rmi.RemoteException; } To compile it: > javac Calculator.java

37. 37 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Implementation of Remote Service public class CalculatorImpl extends java.rmi.server.UnicastRemoteObject implements Calculator { // Implementations must have an explicit constructor // in order to declare the RemoteException exception public CalculatorImpl() throws java.rmi.RemoteException { super(); } public long add(long a, long b) throws java.rmi.RemoteException { return a + b; } …

38. 38 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Stubs and Skeletons To generate the stub and skeleton files, invoke the RMI compiler, rmic The compiler runs on the remote service implementation class file > rmic CalculatorImpl After you run rmic you should find the file CalculatorImpl_Stub.class

39. 39 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Host Server Remote RMI services must be hosted in a server process import java.rmi.Naming; public class CalculatorServer { public CalculatorServer() { try { Calculator c = new CalculatorImpl(); Naming.rebind(" rmi://localhost:1099/CalculatorService ", c); } catch (Exception e) { System.out.println("Trouble: " + e); } } public static void main(String args[]) { new CalculatorServer(); } }

40. 40 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Client import java.rmi.Naming; import java.rmi.RemoteException; import java.net.MalformedURLException; import java.rmi.NotBoundException; public class CalculatorClient { public static void main(String[] args) { try { Calculator c = (Calculator) Naming.lookup( "rmi://localhost/CalculatorService"); System.out.println( c.sub(4, 3) ); System.out.println( c.add(4, 5) ); System.out.println( c.mul(3, 6) ); System.out.println( c.div(9, 3) ); } …

41. 41 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Running the RMI System Make sure you change to the directory that contains the classes you have written Launch a terminal, start Java RMI registry > rmiregistry Launch another terminal, start the server > java CalculatorServer Launch the 3 rd terminal, start the client > java CalculatorClient

42. 42 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Parameter Passing in Java RMI When Java RMI calls involve passing parameters or accepting a return value –How does RMI transfer these between JVMs? –What semantics are used? –Does RMI support pass-by-value or pass-by- reference?

43. 43 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Parameter Passing in Java RMI Primitive parameters: pass by value –For both input parameter and return type, Java RMI makes a copy of a primitive data type and send it to the destination Object parameters: pass by value –The object to be passed, together with all the objects it references, are serialized and copied over

44. 44 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Parameter Passing in Java RMI Remote object parameters: pass by reference –A client can obtain a reference to a remote object through the Java RMI Registry program –A client can obtain a remote reference as a result of making a remote method call

45. 45 Fall Semester 2006EEC-681: Distributed Computing SystemsWenbing Zhao Distributed Garbage Collection in Java RMI Java takes care of memory management Distributed garbage collection is needed in Java RMI –Many challenges to do so. Most prominent problem is that a client might quit without notice –Solution: lease based. The resource is granted to a client for certain period of time. It is the client’s responsibility to renew the lease. The resource is reclaimed if lease is not renewed