Web Services

Denis Helic

IICM, TU Graz

Web as a database

• The Web we use is full of data
  • Book information, opinions, prices, arrival times, misc junk, etc.
• The data is organized around a simple data model: node-link model
• Each node is a data item that has a unique address and a representation
  • Representation is e.g. HTML, PDF,... for humans, or e.g XML for programs
• Nodes can be interlinked using their unique addresses

Web as a platform for distributed systems

• The Web is full of services that allow humans and programs to use the Web data
• Service also have unique addresses
• Use a particular representation for data exchange, e.g. XML, SOAP, WSDL
• Follow a particular architecture that defines how services are used
• You combine a number of services to achieve a desired functionality , e.g. you create a distributed system

Types of services (1)

• What is Google search engine?
  • It is a service for querying a massive database and getting back a formatted response
• What is a Web application?
  • It is a service offering a specific functionality and getting back a formatted response
• What is a Web site?
  • It is a service offering specific information in a specific format

Types of services (2)

• All these services are for users
• However, we are interested in services for programmers
• Such services provide an API
• Programmers use the API, unique addresses, representations of services
• Programmers follow the arch. style to integrate and combine services to achieve a desired functionality
• We will call this part of the Web: the programmable Web

Kind of Things on the Programmable Web

• There exist numerous approaches to web services in all areas
• The programmable Web is based on HTTP and XML for represenation
• Some parts serve HTML, JSON, plain text, binary documents, but most parts use XML
• The only agreement is HTTP - if you do not use HTTP you are not on the Web
• All other things, representations, addresses, APIs differ
• We need a classification!

Classification based on architectural design

• The only common thing: HTTP
• HTTP is a document-based protocol in which a client puts a document in an envelope and sends it to the server
• The server answers by putting a response document into an envelope and sending it to the client

Method information (1)

• Question: how the client conveys its intention to the server?
• How does a server know a certain request is a request to retrieve some data?
• Instead of a request to delete the same data?
• Why should the server do this instead of doing that
• This information is called method information

Method information (2)

• First alternative
• Convey method information by putting it in the HTTP method
• GET, DELETE, PUT, POST
• Semantics is clear
• It is standardized

Method information (3)

• Second alternative: Flickr services
• http://www.flickr.com/services/rest?method=flickr.photos.search&tags=penguin
• http://www.flickr.com/services/rest?method=flickr.photos.addTags&...
• http://www.flickr.com/services/rest?method=flickr.photos.deleteComment&...
• Method information in method query variable, all methods use GET (although some change the data)

Method information (3)

• Third alternative: SOAP and WSDL
• Method information in the SOAP message as defined by WSDL

	wsdl_uri = 'http://api.google.com/GoogleSearch.wsdl'
	driver = SOAP.create_rpc_driver(wsdl_uri)
	result_set = driver.doGoogleSearch(...)

Method information (4)

	POST search/beta2 HTTP/1.1
	...	
	<soap:Envelope>
		<soap:Body>
			<gs:doGoogleSearch>
				<q>REST</q>
			</gs:doGoogleSearch>
		</soap:Body>
	</soap:Envelope>

Method information (5)

	<operation name="doGoogleSearch">
		<input message="typens:doGoogleSearch" />
		<output message="typens:doGoogleSearchResponse" />
	</operation>

• Method infromation defined in WSDL and used in SOAP and via SOAP-framework in the code
• HTTP method is always POST

Scoping information (1)

• Question: How the client tells the server on which data to operate?
• E.g. the server knows that the client wants to delete some data, but which data is it?
• Why should server operate on this data instead of that data?
• This information is called scoping information

Scoping information (2)

• First alternative: as query variables in URL
• http://www.google.com/search?q=REST
• Scoping information is search?q=REST
• Method information is in GET

Scoping information (3)

• Another example Flickr
• Most of the query variables are scoping information
• tags=penguin
• Recollect that Flickr also includes method information in query variables
• Thus, URL contains a method and a number of arguments to that method in the sense of programming languages

Scoping information (4)

• Second alternative: scoping information in the body of a SOAP message

	POST search/beta2 HTTP/1.1
	...	
	<soap:Envelope>
		<soap:Body>
			<gs:doGoogleSearch>
				<q>REST</q>
			</gs:doGoogleSearch>
		</soap:Body>
	</soap:Envelope>

Competing Architectures

• RESTful, Resource-Oriented Architectures
• RPC-Style Architectures
• REST-RPC Hybrid Architectures

RESTful, Resource-Oriented Architectures

• Method information goes into HTTP methods
• Scoping information goes into URL
• More on this architectural style in the next lecture

RPC-Style Architectures (1)

• RPC - Remote Procedure Call
• An RPC style service recieves an envelope full of data from the client
• The service answers with a similar envelope again full of data to the client
• Both method and scoping information are inside of the envelope
• HTTP methods typically POST but sometimes also GET

RPC-Style Architectures (2)

• The best example of the envelope format is SOAP
• There exist other envelope formats like XML-RPC
• Every RPC-style service defines a completely new vocabulary
• e.g. the way how method information and scoping information are represented
• You need another language to define the representation: e.g. WSDL

RPC-Style Architectures (3)

	
	<methodCall>
		<methodName>lookupUPC</methodName>
		<params>
			<gs:param>
				<value>
					<string>0920985</string>
				</value>
			</gs:param>
		</params>
	</methodCall>

RPC-Style Architectures (4)

• XML documents change depending on method and parameters
• Very similar with SOAP (recollect Google example)
• Of course, format is different but the design is the same
• SOAP is used almost exclusively in such an RPC way

REST-RPC Hybrid Architectures (1)

• Services somewhere between RESTful and RPC services
• http://www.flickr.com/services/rest?method=flickr.photos.search&tags=penguin
• Both method and scoping information is in URL
• HTTP is used as an envelope format

REST-RPC Hybrid Architectures (2)

	GET services/rest?method=flickr.photos.search&tags=penguin HTTP/1.1
	Host: www.flickr.com

• Another prominent example of this style is del.icio.us

Big Web Services

• When people think about Web Services they typically think about RPC-style services
• In particular they think about SOAP/WSDL representation of this style (Big Web Services)
• Vendors sell this as a completly new technology or acrhitecture
• But, RPC is there for more than 30 years
• e.g. CORBA, COM, ...

Sample SOAP service (1)

• The service prints out the string it recieves (echo)

public class EchoService {
    public String echo(String msg) {
        return msg;
    }
}

Sample SOAP service (2)

• Invoking the service using HTTP GET

<?xml version="1.0" encoding="UTF-8"?>
<SOAP-ENV:Envelope 
   xmlns:SOAP-ENV="http://schemas.xmlsoap.org/soap/envelope/">
  <SOAP-ENV:Body>
    <echo>
      <msg>Hello World!</msg>
    </echo>
  </SOAP-ENV:Body>
</SOAP-ENV:Envelope>

Sample SOAP service (3)

• Very simple formatting
• echo element is the method information
• msg subelement is the scoping information
• The service guesses the proper datatype conversions
• In this case very simple because the argument and the return value are simple strings

Sample SOAP service (4)

• Response from the service might look like this (most of the namaspaces are left out)

<soapenv:Envelope>
 <soapenv:Body>
  <ns1:echo>
   <ns1:arg0 xsi:type="soapenc:string" 
    xmlns:soapenc="http://schemas.xmlsoap.org/soap/encoding/">
    Hello World!
   </ns1:arg0>
  </ns1:echo>
 </soapenv:Body>
</soapenv:Envelope>

Sample SOAP service (5)

• Additionally, since there is a new vocabulary you need a WSDL
• Even this simple example shows how much overhead is there
• Of course, there are frameworks which create WSDL and SOAP for you
• Still there is processing overhead

SOAP/WSDL Frameworks (1)

• Frameworks provide abstractions of SOAP and WSDL!
• Basically, they allow you to do RPC transparently
• You do not even need to know about SOAP or WSDL
• Frameworks process SOAP and WSDL for you

SOAP/WSDL Frameworks (2)

• .NET from Microsft
• WebSphere from IBM
• J2EE from Sun
• Apache Axis

SOAP and WSDL in Axis (1)

• Java2WSDL tool creates automatically WSDL from Java interfaces
• It creates descriptions of all public methods
• It creates proper datatypes
• It needs to be able to access these datatypes Java classes
• The classes should be JavaBeans

SOAP and WSDL in Axis (1)

• WSDL2Java tool creates Java classes from WSDL descriptions
• Client side
• A Java interface which coresponds to the operations defined in WSDL
• For echo service this would be Echo interface with a single public method

SOAP and WSDL in Axis (2)

• A so-called stub, which is the implementation of the generated interface
• It uses Axis API (e.g., Service, Call) and SOAP to communicate with the service
• A ServiceLocator class which retrieves instances of the generated interface
• Basically it is an Abstract Factory

SOAP and WSDL in Axis (3)

• Server side
• An empty implementation of the generated interface
• You should modify this empty implementation
• Forward the calls to the proper methods of your implementation

Axis - An Example (1)

• PersonDAO interface from the Publication Database
• Basically, we want to offer the access to the database through a service
• The usual way of providing services
• I.e., publishing already existing functionality as Web services

Axis - An Example (2)

• PersonDAO interface

....
public void storePerson(Person person);
public Iterator readAllPersons();
public Person readPersonWithId(int id);
....