|
||||||||||||
So, what does "manage and operate" mean in the context of a smart meter infrastructure? There probably isn't any short answer that will neatly fit in this space, so I'll start with what I'm not talking about. I'm not talking about reading meters, or detecting power outages, or measuring power quality. I'm not even talking about deployment and provisioning; at least, not directly.
I'm talking about network management. Network management includes all of the stuff in between the deployment and the reading of meters. Network management is the care and feeding of the smart meter infrastructure, and if it isn't done well, some or all of the objectives of your smart metering project are designed to fail.
Network Management Fundamentals
Faults include any defect or malfunction that has an operational impact on the network. Faults occur every day, even on the best networks. Left undetected, faults can cripple a network. Faults are generally detected by network management software in one of two ways. First, a node (or nodes) on the infrastructure itself may asynchronously signal that a fault has occurred. Secondly, faults may be triggered by the analysis of historic performance and capacity data.
Configuration includes any soft change to any node on the network. Soft changes include version and configuration. Unfortunately for operators, soft changes to infrastructure are very common. Firmware must be upgraded and configuration parameters must change to enable new business objectives, to fix bugs, to remedy faults, and to optimize performance. Accounting is an odd term for network management, and it is probably more convenient to think of accounting in terms of capacity management. Capacity management is about managing the resources of the network, including memory, transmit and receive buffers, CPU utilization, and bandwidth.
Performance is one of the hardest areas of network management to pin down, but it does mean exactly what it seems to mean. Performance tries to answer the question: How well are things working? In some cases, objective measures are simple to observe (round trip time, application run time, routing convergence, etc.). In most cases, though, performance is a subjective measure that is primarily tied to unique business requirements. Regardless of the measurement, performance management often becomes the dominant aspect of network management.
Security is the final broad area of network management. At its most basic, security management includes alerts based on clearly identifiable security breaches or attempted breaches. For example, a security alert may result from an unauthorized network device attempting to send routing updates. Security management can also include indications of abuse or excessive use of resources. Security concerns are so ubiquitous in network management that they impact every other management category.
Network Management for Utilities
Fortunately for all of us, networks and network applications have come a long way, and we're finally seriously talking about networking public utility infrastructure. Networking at the scale of a utility is not easily accomplished, though, not even with current technology, and it presents new network management challenges in terms of scale, user interface, and automation.
Smart meter networks are taking network scale to levels previously unthinkable. A single smart meter deployment may have anywhere from 100,000 to more than 10 million end-points and intermediate nodes. In some architectures, each end-point may also serve as a router. For the sake of comparison, even the largest Internet Service Providers typically manage far fewer than 100,000 routers within their infrastructure -- and that's on the small end of a smart meter deployment.
Regardless of the total number, each device in a smart meter network must be managed. Each device has software that must be upgraded periodically. Each device has a configuration that must be managed. Each device has resources that must be used effectively, and the network at large has a finite amount of bandwidth available across all of those devices. Each device must adhere to overall performance and security policies, as well. In addition to the sheer number of devices that utilities will be operating, they will also need to cope with the massive accumulation of management data over time.
Just as network management data may overwhelm disk arrays, it may also overwhelm operators. Network management system user interfaces typically represent networks with charts, graphs, event lists, and logical maps. There's nothing necessarily different in the approach that a smart meter operation may take, but the requirement to filter and distill information from management data has never been more important. Also, due to smart meter vendors' use of radio frequency and mesh routing technologies, logical maps will have to be augmented by accurate geographical representations of the infrastructure.
Automations must be ubiquitous in order for massive smart meter networks to be successful. Network software and configuration upgrades alone could threaten the viability of a project if not sufficiently automated. Automations around capacity, performance, fault, and security policies will also be necessary in order to maintain an effective network infrastructure. Long gone are the days of troubleshooting every slow network path with a manual traceroute or poking around a configuration from a command line. These new networks are just too big for that, and we'll need good management software and new techniques to bridge the gap.
Bridging the Gap
One approach to address each of the challenges of managing a large smart meter network is to employ a network element management (NEM) solution compatible with the rest of your smart grid communications applications. Combined with a compatible distributed networking technology and data stores, this approach allows you the ability to manage your network elements in the following ways:
Fault detection and reporting will be handled by standard asynchronous trap messages sent to the network element manager from any compatible communication device. Faults may also be detected based on performance or capacity policies that are applied to periodically polled metrics. Faults can even be detected from some other applications. All faults are represented in the NEM event list, and can even be filtered and displayed geographically. Faults may also be forwarded (also by policy) to a third-party manager of managers (MoM). Event forwarding allows operators to deal with faults across a heterogeneous infrastructure from a single MoM console.
Capacity (or Accounting, if you'd prefer) metrics will be periodically gathered from all infrastructure devices. A set of standard reports is provided for the most important capacity metrics, while any capacity metric (or set of metrics) may be charted on demand. NEM will also automatically manage capacity against thresholds applied to any metric. Whenever a threshold is violated an event is created in the event list, which can also be forwarded to a MoM if so desired.
Performance metrics are managed in much the same way as capacity metrics. Some of the standard performance metrics measured by NEM should include routing efficiency, network stability, and network availability. Standard reports should be available for the most important performance metrics as well as on-demand charts. Routing metrics may even be displayed geographically, thus allowing operators to easily identify environmental issues and "pockets of badness" that would otherwise be hidden in the network.
Security management will be ubiquitous throughout technologies in the system. Applications are secured with user roles, credentials, and encryption. Network resources are secured by key exchange for authentication and encryption for privacy. Device security is managed by proactive alerts (traps) sent to NEM and security counters maintained on each device.
NEM will also provide an interactive diagnostic console for those times when manual troubleshooting can't be avoided and a network management dashboard for those times when only a glance is necessary.
Conclusion
Network management should be a major consideration when evaluating and deploying a smart meter network. Smart meter projects will result in some of the largest networks the planet has ever witnessed. Those networks surely cannot succeed without management solutions that are capable of addressing all of the major management categories while also meeting the new challenges presented by tremendous scale and new technologies.
