| The market for 2.5G and 3G services is taking shape | | | | the performance of applications before deployment |
| as mobile providers offer customers capabilities that | | | | may make success itself a risky proposition. After all, |
| include Internet browsing, chat, multimedia downloads | | | | an application that performs perfectly well with |
| and online purchasing. These services promise to | | | | several hundred simultaneous users may not survive |
| generate significant incremental revenue for | | | | the onslaught of tens of thousands. So, without |
| providers, as customers embrace the next-generation | | | | appropriate preparation, providers can find |
| mobile experience. They will also be important to | | | | themselves unable to cope with the very success |
| other market beneficiaries - including content | | | | they were hoping for! |
| providers and online merchants - who see the | | | | The Solution: Network-Aware Development |
| affluent mobile user as an important target. | | | | One solution to this problem is to make the |
| To ensure the success of their 2.5G and 3G services, | | | | real-world conditions of the production network a |
| however, mobile providers have to grapple with | | | | central design consideration for all applications from |
| three primary risk factors. First, they obviously have | | | | the earliest stages of design. In other words, to |
| to ensure that the services they offer are the ones | | | | ensure service quality in the real world AFTER |
| buyers want. Only by understanding customers' | | | | deployment, it is essential to be able to discover any |
| wants and needs can providers offer a portfolio of | | | | potential performance problems with an application's |
| services that will generate the necessary take-rates | | | | behavior in the real world BEFORE deployment. |
| to be profitable. Second, they have to design those | | | | This can only be done, however, if developers have |
| services to appeal to users despite small screens, | | | | some practical means of modeling those conditions in |
| slower access speeds and limited keypad functions. | | | | the lab. Ideally, such a modeling environment will be |
| These design skills are essential for optimizing the | | | | able to replicate conditions on the production |
| user experience and market acceptance. | | | | network - including bandwidth constrains, the number |
| Third, providers must ensure the reliability of these | | | | of distribution of end-users, etc. If this modeling |
| services. It's one thing to tolerate a temporary | | | | environment is then connected to the actual server |
| outage on a free or bundled service. It's quite | | | | infrastructure that will support the planned |
| another thing to lose access to or experience | | | | applications, the behavior of those applications in the |
| chronically sub-par performance with a service you're | | | | real world can be accurately observed and tested. A |
| paying for. That's why it's absolutely critical for mobile | | | | variety of "what-if" scenarios can also be generated |
| providers to validate the performance of their 2.5G | | | | and observed to determine their impact on service |
| and 3G services before they roll them out to | | | | levels: the loss of some portion of the network, an |
| customers. Application failures are something mobile | | | | exceptional peak in service utilization, a spike in other |
| customers are unlikely to be very forgiving about. In | | | | types of network traffic, and so forth. |
| fact, too many failures too early in the game may | | | | In addition to accurately replicating existing and |
| permanently turn customers off to these | | | | projected real-world network conditions, an effective |
| money-making services-dooming the market before it | | | | modeling environment should also lend itself to |
| has a chance to take hold. | | | | collaboration between application development teams |
| The Service Level Assurance Challenge | | | | and network infrastructure managers. By sharing a |
| Ensuring service levels for 2.5G and 3G services | | | | common modeling platform, these two groups can |
| presents special challenges. Multimedia applications, for | | | | reach consensus about issues potentially affecting |
| example, are particularly bandwidth-hungry and are | | | | service levels - as well as the optimal remedy for |
| easily prone to session breakdowns due to | | | | such issues - rather than getting bogged down in |
| temporary congestion in the network. In addition, | | | | conflict and finger-pointing. A multi-purpose modeling |
| such applications often use multiple session, signaling | | | | environment also ensures that any investment in |
| and data protocols. These complex dependencies | | | | modeling technology can be fully leveraged across |
| often make it difficult to maintain service levels and | | | | the entire application lifecycle, including change |
| understand the issues that may compromise the | | | | management, capacity planning and CapEx projection. |
| end-user experience. | | | | Ensure Risk-Free Wireless Deployment |
| In fact, providers often find themselves rolling new | | | | No provider can afford to have their early wireless |
| services out into production without fully | | | | service efforts tarnished with chronic poor |
| understanding how those services will really be | | | | performance. And no one wants to surrender their |
| experienced by customers under real-world | | | | hard-earned early adopter customer to the |
| conditions. That's because services are typically | | | | competition because of service failures. Mobile |
| tested in laboratory environments that don't | | | | providers who want to gain and sustain marketshare |
| adequately reflect the bandwidth limitations, | | | | in value-added wireless services must therefore |
| distance-driven latencies, and capacity contention that | | | | make every effort to ensure the reliability and |
| exist in the production environment. This is | | | | performance of their offerings. And to do so, it is |
| unfortunate, because it exposes providers to the | | | | critical that they embrace modeling technologies that |
| considerable risk that applications won't perform in | | | | are as advanced and sophisticated as their |
| production as they did in the lab. | | | | next-generation networks. |
| By the same token, the inability to adequately assess | | | | |