Technical Article Mobile versus Mobile: An Identity Crisis January 16, 2012 by Lori MacVittie 3341 article acceleration apm application delivery applications architecture availability context-aware design dev devops dynamic infrastructure hardware html5 infrastructure irules management mobile network-side scripting performance security service provider strategic point of control us web 2.0 0 #mobileThe expansive options consumers revel in creates an identity crisis for IT that is best resolved via context-aware mobile mediation. Back in the days of the browser wars, when standards were still largely ignored and the battle for the desktop was highly competitive, developers had to make choices and compromises. They could either write extensive client-side scripts to detect the user’s browser and address the peculiarities of that environment or they could simply ignore them with a disclaimer that “this site (works best when viewed in | was written for) browser X.” As time went by, developers were able to discontinue this annoying practice as browser features converged and a common, standardized platform emerged upon which all applications were able to be delivered to any popular browser without concern. Then mobile phones appeared, and the user experience degraded again, this time driven by the relatively feeble processing power of the platform. Small screens aside, the memory and processing power available on a mobile phone was such that – when combined with a constrained networking environment – the delivery of increasingly chunky, graphic heavy, interactive applications to mobile phones was simply a bad idea for both organizations and its visitors. Developers and web ops returned to the inspection of HTTP headers to determine whether or not a visitor was using a mobile platform, and began writing leaner, more compact interfaces specifically for those platforms. Enter tablets. Neither desktop nor phone, these admittedly mobile platforms are compact but nearly as powerful as their overweight tethered cousins without sacrificing the mobility of their anorexic form-factor brethren. COMMON GROUND: MOBILE OPERATING SYSTEM Tablets are certainly able to take advantage of emerging web application standards such as HTML5 to deliver a full, rich interactive desktop-style experience on a mobile platform, but are rarely offered it. They are by default offered up a mobile experience because of a common element: the operating system. Even if the operating system was masked, still they’ll find out by checking a second, lesser known HTTP header: HTTP_X_WAP_PROFILE. And if not there, then potentially in HTTP_PROFILE. Unfortunately, this is not necessarily the fault of the developer. There are no standards prescribing what should or should not identity a mobile device, and thus manufacturers are left to their own devices (pun only somewhat intended). There is no reliable way, today, to identify a mobile device accurately. Developers are left writing complex scripts that strip apart user agents, profile strings and whatever other contextual data they can extract from HTTP headers to determine how best to serve up content. That often means visitors are served content that is not wholly appropriate for their device. Even if they could, the market is so volatile at this point that it’s a sure bet that a new device will enter soon that requires modification to the application yet again. More code, more string manipulation, more latency in processing. CONTEXT-AWARE MOBILE MEDIATION It would be nice if developers could simply receive accurate inbound HTTP headers. Headers that clearly identify the device not only from an operating system perspective, but from a form-factor and network perspective in a standard way. But there are no standards, yet, and may never be. Thus a solution may be found in imposing standards upon inbound requests specifically for developers to better address the disparities in resolution, functionality, and performance between the various mobile device types. This requires some amount of pre-planning. Design, if you will, or architecture up front. It requires that developers and devops sit down together and determine a standard means of communicating information between infrastructure and the applications it supports. Consider the possibility of two custom HTTP headers, one identifying the network type and one specifying form-factor and device: HTTP_X_NETWORK = “WIFI | MOBILE | LAN” HTTP_X_DEVICE = “TABLET | PHONE | DESKTOP” If developers could rely on these two custom HTTP headers to exist for every inbound request, they could then develop applications based upon these characteristics that were more appropriate for the given device and network over which the device connected. Implementation requires only minimal inspection and insertion on a context-aware mediating device such as a network-side scripting capable application delivery controller. Because the application delivery controller is topologically positioned in a strategic point of control, it has visibility into the network, client, and server-side environments. This gives it the ability to better interpret and execute policies that govern the delivery of applications to optimize performance and assure availability, but it also provides the ability to extract and share pertinent data with applications and other infrastructure. This data can be shared in a number of ways, including modification of the payload, of the headers, insertion of new headers, removal of old headers, etc… The flexibility inherent in network-side scripting solutions, particularly those capable of side-band connectivity, allows devops and developers to design and develop a solution that works for them – in their environment. The advantage to such a solution lies not only in more accurate, actionable data to share with applications, but in its ability to easily be modified without negatively impacting the application. A second advantage is the ability of developers to also take into consideration the network characteristics of the mobile device, data generally not available or, if available, generally inaccurate. A mobile device today may be accessing an application via WiFi or a mobile network, and that piece of information is quite pertinent as the performance and capabilities of each network are quite different and have a significant impact on the end-user experience from a delivery perspective. Yet this data is not available by default to developers and it cannot reliably be inferred from device type. By leveraging a context-aware mediating solution, however, it becomes possible to share this data with developers such that they are able to take that information into consideration when putting together a response to a given request. While not a panacea, such a solution certainly provides a more consistent and overall accurate environment in which to deliver applications to the increasingly broad and diverse spectrum of mobile devices. Stack Overflow: How do detect Android Tablets in general. Useragent? Mobile Browsing Reaches All Time High The Magic of Mobile Cloud Understanding network-side scripting At the Intersection of Cloud and Control… Cloud-Tiered Architectural Models are Bad Except When They Aren’t WILS: WPO versus FEO Fire and Ice, Silk and Chrome, SPDY and HTTP Grokking the Goodness of MapReduce and SPDY last modified: January 16, 2012 0 Comment(s): You must be logged in to post comments.