Client-Side Storage in Web Applications

More documents

Recommendations

Info

By using a client-side storage mechanism, most of the aforementioned disadvantagescan be overcome. Figure 2 illustrates the modern interaction model, which includesa client-side storage. First, a web page checks whether the client-side storagealready has the requested new data. If yes, then the data is immediately displayed onthe user interface without communicating with the server. Otherwise, the requestednew data is retrieved from a database residing on the server, and when retrieved, thedata is stored in a client-side storage for future use. This approach not only makeswebsites faster, but it also enables offline support and reduces server load.Figure 2. Modern interaction model.HTTP cookie [3] was the first mechanism, which enabled web applications to storedata on web clients. However, cookies do not meet the requirements of modern webapplications. Thus, new client-side storage mechanisms were needed. First came(Google) Gears [7], and then influenced by it, emerged various W3C-specified APIproposals [10, 11, 17]. Other useful and closely related W3C-specifications includefull Offline Web Applications [9], File API [19], and Web Workers [12]. However,these specifications are out of the scope of this paper.The rest of the paper is organized as follows. Section 2 examines very briefly thehistory of client-side storage. In Section 3, recently specified W3C API proposals forstoring and manipulating persistent data on the client side are presented in detail. Nextsection introduces research projects, which utilize these modern client-side storagemechanisms. Finally, conclusions are drawn in Section 5.2 EvolutionThis section presents the evolution of legacy client-side storage mechanisms. Thesection starts by giving a very brief overview of the three popular storage mechanisms:HTTP cookie [3], Flash cookie [2], and (Google) Gears [7]. In the end, thefeatures of these three mechanisms are compared against a set of evaluation criteria inorder to highlight the similarities and differences of each mechanism.2
2.1 HTTP CookieHTTP cookie 1 [3], invented by Netscape in 1994 and later standardized by IETF, isa popular mechanism that lets web servers and web clients to maintain and manage astateful session between each other over the mostly stateless HTTP protocol. Cookiesallow a small piece of simple key-value pair data (4 kB per cookie, 50 cookies perorigin) to be stored in a user’s web client (browser). Cookie data can be set to lastover a single session (session cookie) or multiple sessions (persistent cookie). Typically,web applications use them for session management, authentication purposes,storing user preferences, and user tracking. All major browsers have native supportfor cookies.Although cookies provide a simple and well-supported storing mechanism, theysuffer from several problems. First, each cookie is sent back and forth with everyHTTP request (via HTTP headers), which adds a lot of unnecessary overhead. Second,their storage size is way too small for modern web applications. Third, the sameweb application using cookies cannot be run in multiple web browser tabs simultaneously.Finally, the API is a bit clunky. [15, 16]2.2 Flash CookieIn 2002, Adobe introduced a feature called Flash cookie (Local Shared Object) [2],which is an Adobe Flash Player based mechanism to store and manage data on auser’s computer. Flash cookies are exactly like HTTP cookies, except that they canstore more complex data types than just plain strings and offer more storage space(defaults to 100 kB per origin). The easiest way to access Flash cookies fromJavaScript is via the ExternalInterface object, which was introduced in Adobe Flash 8in 2006.In order to use Flash cookies, Adobe Flash Player 6+ plug-in needs to be installedto a web browser. Mobile browsers, however, are gradually moving away from Flash,and therefore the use of Flash cookies is no longer an option on mobile devices [1].2.3 (Google) GearsIn 2007, Google launched Gears 2 [7] which offers web applications, among otherfeatures, the ability to store and manage persistent data locally on web clients, andaccess to it through the use of a JavaScript API and a variant of SQL. Gears comeswith an embedded client-side relational database based on SQLite 3 . Compared toprevious solutions, Gears’ main advantage is its unlimited storage size.In March 2011, Gears announced that there will be no new Gears releases in the futureand newer browser versions will not be supported [8]. The decision was based onshifting the focus from Gears to HTML5, i.e., to help the W3C in defining and stan-1 Also referred to as a cookie, Web cookie, and browser cookie.2 Formerly Google Gears3 SQLite, http://www.sqlite.org/3
Page 1: Client-Side Storage in Web Applicat
Page 5 and 6: common objective shared by all the
Page 7 and 8: 5: var myData = {6: "code": "T-111.
Page 9 and 10: 6: "code": "T-111.5502",7: "name":
Page 11 and 12: storage should be trusted, as a mal
Page 13 and 14: Silo was evaluated with multiple re

Client-Side Storage in Web Applications

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?