Crud Service for Client and Server Timing Computation using HTTP Methods

DOI : 10.17577/IJERTV10IS070022

Download Full-Text PDF Cite this Publication

Text Only Version

Crud Service for Client and Server Timing Computation using HTTP Methods

Manish L. Jivtode

Department of Computer Science,

Janata Mahavidyalaya, Chandrapur, Maharashtra, India

Abstract:- With the increasing popularity of the cloud, web service technology has emerged as a popular way for building distributed applications involving distributed databases. The sensitive data exchanged between web services and their clients hosted in the cloud data centers must be protected while in the transit. The distributed databases and applications in these geographically dispersed data centers belong to either internet or intranet domains.

In this paper, study of weakness and suggest improvements of CRUD service contract and a set of operations like CREATE, RETRIEVE, UPDATE, and DELETE (CRUD) by using HTTP verbs like POST, GET.

Keywords: Web Service, Distributed database, REST web Services, CRUD service, GET method and POST method.

  1. INTRODUCTION

    Recent improvement of web services technology provide uninterrupted, lightweight, continuous web services to resource constrained devices like mobile, laptop, desktop, tablet etc. in wireless or wired environment. The services are based on service oriented principles and follows four tenets of service oriented development

    1. Boundaries are explicit,

    2. Services are autonomous,

    3. Services share schema and contract, not class,

    4. Service compatibility is determined based on policy.

    Web services propose a service-oriented paradigm for computing in which distributed, loosely coupled services collaboratively implement business processes and accessed via the Internet by end users. The use of Web services over the intranet and internet has increased rapidly. Web services are used to support application-to-application communication and to address interoperability issue for systems integration project. These Web services provide a standard-based approach for different software applications or components involved in supporting real-time information retrieval or presenting dynamic context-driven information to the user. Web services rely on a set of standards to support interoperability among applications developed in different languages and running on different platforms or operating systems.

    Core web services standards including SOAP (Simple Object Access Protocol), WSDL (Web Services Description Language), and UDDI (Universal Description, Discovery, and Integration).

    Figure 1.1 Web Service Model

    (Source: Borrowed from [2])

    Web Services are considered an instance of an SOA. The web services network is an application level network involving a number of participants- providers, service consumers, and service registry operators.

  2. REPRESENTATIONAL STATE TRANSFER (REST)

    REST is architecture for developing web services. REST architecture uses HTTP or similar protocols, by constraining the interface to a set of well-known standard operations (i.e., GET, PUT, POST, and DELETE for HTTP). REST architecture is designed to show how existing HTTP is enough to build a Web service and to show its scalability [1]. Roy Fielding, who is also one of the principal authors of HTTP, in his doctoral thesis [7], Architectural Styles and the Design of Network-Based Software Architectures, defined a set of architectural principles for the Web called REST. REST architectural principles can be used for designing Web service solutions. REST focuses on a systems resources, including how resource states are addressed and transferred over HTTP [4]. The main idea behind REST was to use well-developed HTTP for transferring data between machines, rather than using a protocol that works on top of the HTTP layer for message transfers. An application designed following REST principles would use HTTP to make calls between the machines, rather than relying on complex mechanisms like CORBA (Common Object Request Broker Architecture), RPC (Remote Procedure Call), or SOAP. Therefore, REST applications use HTTP request functions to post data, read data, and delete data, thus using the full functionality of HTTP Create, Read, Update, and Delete (CRUD) operations. REST can run on HTTPS, providing for the secured transmission of data. The CRUD operations in conjunction with HTTP REST functions are shown

    Table 1.1 HTTP methods and CRUD action

    CRUD operation

    REST keywords (HTTP)

    READ-read, retrieve

    GET

    CREATE-create or add new entries

    POST

    UPDATE-update or edit existing data

    PUT

    DELETE-delete existing data

    DELETE

    (Source: Borrowed from [6])

    The application of REST principles for web services requires HTTP protocols find it easy to understand and apply REST principles. The term RESTful is like object-oriented, a language, a framework, or an application that may be designed in an object-oriented way, but that does not make its architecture the object-oriented architecture [5]. RESTful Web services are emerging as the choice for many of the leading Internet companies to expose their internal data and functionalities as URI identified resources.

  3. EXPERIMENTAL WORK

    1. DATA SET USED

      The availability of the standard data set in the field of security in distributed database is limited, six of the standard dataset collections were considered for the study and performance evaluation. The basis for selection of these datasets is availability for query and relevance judgments with these datasets. These collections are described as follows:

      Table 1.2 Data sets used for performance evaluation

      Name

      Description

      Data Size

      Number of queries tested

      GET

      POST

      PUT

      DELETE

      Text

      Plain text

      1KB 16KB

      10

      10

      10

      10

      Photo

      JPEG Image

      1KB 16KB

      10

      10

      10

      10

      Audio

      MP3

      1KB 16KB

      10

      10

      10

      10

      Video

      WMV

      1KB 16KB

      10

      10

      10

      10

      Headers

      Custom headers

      Few bytes

      10

      10

      10

      10

      Message

      Message body

      1KB 16KB

      10

      10

      10

      10

      (Source: Compiled by Researcher)

      Four standard REST web service methods were tested for analyzing performance of the implemented system. They include GET, and POST methods. GET method does not carry payload in the message body in request part and returns data in the response only. The data is enclosed in XML tags and placed in message body of POST requests.

      In GET request, an unencrypted plain text request along with custom header (for authentication) is sent to the REST web service and HTTP message body does not contain any text/image/audio/video data. Upon reaching HTTP GET message at REST web service, headers areprocessed and appropriate resource is returned in encrypted form during GET response. The received data is decrypted at the client side and are verified.

      In POST requests, data is signed and then encrypted before sending it through the POST request message body. Signed values of data are asymmetrically encrypted and symmetric key sent through custom headers. In HTTP, message reaches REST service, it is first decrypted using custom header data and then verified by applying encryption/decryption algorithm at the REST web service extensibility points. The messages processed by REST web service by creating a new resource on the REST service database and signed and encrypted response sent back to the REST client. Upon receiving signed and encrypted response at the REST client, it is decrypted and verified and finally presented at the client side. Image/audio/video data is converted to Base64 binary or Base64 string before sending to the REST web service.

    2. CLIENT AND SERVER REQUEST / RESPONSE TIMING COMPUTATIONS

      GET request sent from REST client, it does not carry any payload and only plain text request sent to the REST service. In GET, client authentication information sent through custom headers. It returns an encrypted response from REST web service. Both REST web service and REST clients are programmed to compute the encryption, decryption timing. GET response return text, JPEG image, audio or video data from REST web service to client. Request timings are calculated by computing time elapsed between point A and point B. GET request may result in retrieving a resource from the local database on the client side.

      Figure 1.2 Request/Response Timing for Client and Server

      The experiments were performed to record signing, encryption, decryption, verification, request and response timing for various datasets by programming WCF REST web service extensibility points.

      GET request/response timings are calculated by computing time elapsed between point A and point B. GET request/response results in retrieving a resource from the local database.

      Points A B -> GET request Time T1 = DateTime.Now()

      //Code for sending request Time T2 = DateTime.Now()

      Request timing during GET = T2 T1 Points D, E, F, G -> GET response

      Response time = (T2 T1) + (T3 T2) + Time from E to F +(T5 T4)

      + (T6 T5)

      Table 1.3 Client and Serve Request/ Response Time

      File size

      Request time in milliseconds

      Response time in milliseconds

      Client Side

      Server Side

      Server Side

      Client Side

      1KB

      330

      203

      54

      1232

      90

      59

      2KB

      380

      255

      71

      1249

      109

      67

      4KB

      394

      329

      98

      1278

      114

      86

      8KB

      445

      439

      113

      1297

      169

      124

      16KB

      586

      474

      167

      1311

      187

      199

      (Source: Compiled by Researcher)

      Request / Response Time for Clint and Server in milliseconds

      This graph shows the performance measurement of encryption/decryption of various data sizes versus time and clearly indicates that as the data size increases from 1KB to 16KB, the encryption and decryption time also increases.

  4. CONCLUSION

    From the study, it is concluded that, the size of data is directly proportional to the time that is Fs Te and Fs Td, where Fs is the file size. It indicates that Fs = k1Te, then k1= Fs / Te and Fs = k2Td, then k2 = Fs / Td, where k1 and k2 are the proportionality constant indicates that the rate of change of request time and response time with respect to size of data transferred. GET method scores higher performance than others. Text data set shows best performance compared to image, audio and video data sets since it is uncompressed plain text in original form.

  5. REFERENCES

    1. A. Pironti, N. Mavrogiannopoulos."Length Hiding padding for the transport layer security protocol". Internet draft-pironti-tls-length-hinding-00, IETE Secretariat, February 2013.

    2. A. Barth, C. Jackson, and J. C. Mitchell. Robust defenses for cross-site request forgery. In CCS08: Proceedings of the 15th ACM conference on Computer and communications security, page no. 75-88, 2012.

    3. Dropbox model. "REST Application Programming Interface". 2012.

    4. Erik Christensen, Francisco Curbera, Greg Meredith, and Sanjiva Weerawarana. WSDL 1.1 Document. March 7, 2011.

    5. A. Langley. Unfortunate current practices for HTTP over TLS, 2011.

    6. A. Sotirakopoulos, K. Hawkey, and K. Beznosov. On the challenges in usable security lab studies: lessons learned from replicating a study on SSL warnings. In Proc. of SOUPS, page no 3-18, 2011.

    7. F. Lascelles. "RESTful web services and signatures". October 2010.

    8. Aberdeen Group, Federated Identity Systems An Executive White Paper, Aberdeen Group Inc., Boston, page no 11, 2007.

Leave a Reply