|
|
article page
| 1 | 2 | 3 |
revenues, while decreasing costs of IT firefights.
Doing all of this is not simple; however the best practices of innovative operators provide some guidance. Alltel (now Verizon, US) relies on proactive APM to manage their highly successful “My Circle” service. It allows subscribers free calling to 10 numbers that can be updated at any time only through a Web-based interface. SingTel Optus (Australia) uses APM with their Zoo SDP to manage the real-time revenue stream between subscribers and content providers. Looking at over one-hundred communication service providers world-wide who have deployed APM to manage their next generation services and self-service, we identified some key requirements for the modern day service assurance:
- Deep diagnostics and visibility into the execution environment in production: portals, partner gateways, Java or .NET application servers, SIP/JSLEE, Web services and SOA, all the way to network interfaces (Parlay, Parlay/X). The root cause of a problem can be found in any component, from applications and infrastructure to objects and methods. You need to be able to measure each of these components’ response times, number of invocations, and generated errors. Deep visibility helps trace individual user transactions through different components and systems, and correlates them with other events. Only this will give you a complete picture of your application’s health.
- Monitor all data all the time to catch early symptoms as they happen. Traditionally, interval probing is used to monitor networks. Networks’ topology rarely changes, and most configurations are known in advance. It is possible to design sufficient probes to test the network and detect any outages. With software applications, probing is not good enough. The number of logical paths that the execution process can take is much higher, and is much more prone to changes with software updates. Also, software problems may happen under specific sets of conditions involving usage patterns, load, and other factors. With interval probing, these problems may fall between the intervals or not manifest themselves with synthetic transactions. This is why you need to monitor real user transactions, in real time. Probing still makes sense under light load, but it will not tell how the system would behave under the real load. Also, ALL data needs to be monitored for deep analysis.
|
|
The ultimate goal of service assurance is to prevent problems or detect them before customers are impacted. |
|
- Management should be non-intrusive and have little or no impact on the managed environment itself. In most cases, solutions suitable for development or quality assurance environment are not applicable in production due to their high overhead. Today there are ways to optimize metrics collection so overall additional load on the system is minimal.
- Role-based customizable reporting is essential to give different roles within your organization (business managers, NOC and IT administrators, or QA and software engineers) the appropriate view relevant to their function. While the source of the performance information is the same (coming from your APM solution), the usage requirements and the actions taken are very different. This relates to all the dashboards, reports, and alarms that your APM solution generates.
The bottom line: as telcos are increasingly using standards-based software for new applications and services, managing performance, availability, and the overall customer experience is becoming a new challenge. Application performance management tools, when used proactively as part of SDP and OSS/BSS architectures, can ensure high quality of service, low costs, and top customer satisfaction, ultimately making operators successful in this new environment.
article page
| 1 | 2 | 3 | |
|
|
|
