|
|
article page
| 1
| 2
| 3
| 4
| 5
| 6
| 7
| 8
|
management defy the traditional agent/manager device management model. They involve interactions of users, many networks, and many kinds of service providers.
Classical device management comes in two broad flavors: the service provider ITU/TMForum OmniPoint pyramid and the IETF derived enterprise device management market. Service providers follow a network element (E) that has an on-board active element agent (EA) that connects to element manager (EM) that connects to a network manager (NM) that connects to manager of managers (MoM), which, these days, connects to a service manager (SM), which often involves significant process automation. Each layer isolates the layer below, filtering information up and commands going down and providing some layer autonomy in its processes and data. Many communications protocols with devices are supported including CMIP, TL1, SNMP, CLI, & HTML; yet all of these simply pass data and commands via a 'hands off' management style. Modern mediation products allow fast transactions and scaling of information capture.
Enterprise Device management architecture and implementation is simpler. A passive agent with an onboard MIB connects via SNMP (or HTML) to a manager that connects to a console. To address scaling issues, satellite management services, such as data collectors, feed the manager. Over time, vendors blurred the implementation boundaries between the enterprise market and the service provider market. However, in every case, the network manager sits in the middle chatting via inter-mediators with all the devices. This worked somewhat with relatively small numbers (millions) of stupid devices.
While some members of the TMForum group are exploring new approaches, others continue to adhere to traditional management viewpoints. Both perspectives describe mesh networks and domains of managed devices, with everything eventually connecting upward to a network element management controller.
Swimming Up to Our Neck in Services
Traditional device management frameworks face a significant, current hurdle. In today's world, devices are smarter – they can download and host services. Once this was the realm of personal computers; unfortunately, in the many decades of computer use, no significant management standard was developed to manage smart, flexible devices with interchangeable services. Small pieces were standardized, such as with the DMTF information model, but basically this is a Wild West zone of proprietary solutions.
|
|
If the user's smart phone participates in a denial of service attack, is the user legally responsible?
|
|
In the future of device management, more than physical devices must be managed; the services must also be managed. Sometimes the service is on the phone via manufacturer-installed software. Sometimes, as with the trend in smart phones, these services are downloaded by the user to become resident on the phone. Sometimes the services are in the network such as traditional call handling and messaging; and other services are created and hosted by OTT service providers.
Today, the service/device interaction zone is the domain of mobile phones and computer. Tomorrow, many devices may have these capacities. So our concept of a unified device management framework must encompass devices and services.
Yet, services are managed by a different set of approaches, architectures, and products than are devices. Mostly today, the standards for service management are the work of W3C and web services. More pervasive so far, however, are the many strong proprietary approaches that exist, most notably from IBM and Microsoft. Sun Microsystems' Jini provided a mechanism for the MCI NewWave approach of a specific service management interface for services that complemented the functional service interface. Microsoft seems to be leaning toward this approach with perhaps the most modern service management approach today: CSF.
Another layer of complication is added when you consider the topology and grouping of service and device. Services can be vertical applications interlinked to traditional telco service providers tied to devices. Services can come from OTT service providers and link to edge hosting platforms, such as a web service on Google. Services can interact in meshed peer groups such as the aforementioned icrowdsurf. There are devices/services which
interact with and cluster around a local laptop/server. The intelligent home is an example of a type of device clustering. Sensor networks (industrial, traffic, etc) are yet another type of clustering. Clusters of devices might even be associated with organizational and geographical regions, such as an office or academic campus. Electric, water, and other utility networks with their sensors, switches, controllers, and data gateways form a device cluster that again complicates device management, this time adding issues of national security to the device management problem.
article page
| 1
| 2
| 3
| 4
| 5
| 6
| 7
| 8
| |
|
|
|
