How to hook up a new AMM White Paper
Transcription
How to hook up a new AMM White Paper
White Paper How to hook up a new AMM system to your IT infrastructure Scaling the Gridstream HES for utilities with up to 3 million metering points White Paper Executive summary 3 1. H ow to build multi-vendor systems and optimize the AMM system configuration for up to 3 million metering points 4 2. Understand the big picture 4 3. How to integrate easily? 5 4. Proven platform for optimizing system HW performance and costs 6 4.1 Test setup description 6 4.2 Test results description 8 4.3 Conclusions drawn from the test 9 How to hook up a new AMM system to your IT infrastructure 2 White Paper Executive summary With only three servers, the Gridstream HES test system for smart metering data collection and delivery demonstrated the ability to handle the test profile data sets for up to three million metering points well within a day. The HES also delivered the first test profile data, i.e. three day registers and one hourly register, well within a typical eight hour data SLA. This ensures that the smart metering data is ready for the utility’s processes and operations for the first office hour on a daily basis. Gridstream HES’ substantial data delivery capacity with a decentralized architecture allows us to optimize the system for the diverse needs of utilities. As a result, users can implement various strategies to optimize system cost, reliability and performance. Gridstream HES provides a smart metering data collection platform which serves utility processes and IT systems, and it includes efficient tools to manage the meter park. The system uses an open standard based integration layer that secures future system compatibility. An established integration process reduces the utilities’ and integrators’ effort and risk during integration projects. Finally, the head end system architecture has proven its ability to provide both performance and adaptability for small and medium size systems up to millions of metering points. The flexible architecture in combination with Landis+Gyr’s expertise is available for IT managers to optimize system performance, cost and maintainability. How to hook up a new AMM system to your IT infrastructure 3 White Paper 1. How to build multi-vendor systems and optimize the AMM system configuration for up to 3 million metering points Distribution Network Operator utilities (DNOs) which start implementing smart metering are dealing with diverse IT systems used for billing, balance settlement, customer service, on-site work order management, network operations and network planning. Typically, these systems are put to use over a long time period and they are provided by several different vendors. Utility IT managers are facing the challenge of integrating these systems with smart metering systems to support utility processes, ensuring required overall performance and managing responsibilities between different system vendors. In this white paper, we consider three key areas to focus on when building the AMM IT infrastructure: 1. understanding the role and purpose of each system layer so that they support the utility’s current processes and enable flexibility for the future 2. managing the integrations between systems, typically provided by several vendors, and 3. ensuring that the AMM data collection system setup is performance optimized and so minimizes overall costs. We will present the Gridstream HES system, which enables utilities to build AMM system networks that are adaptable for future needs, and helps to manage system integrations with less risk and clear responsibilities between the integration parties. Finally, we will use the test results to show how the Gridstream HES architecture allows system setup optimization for different sizes of AMM data collection systems up to 3.000.000 metering points. 2. Understand the big picture DNO utilities have quite similar business goals but may use very different types of resources and processes to reach them. This has led to very diverse IT infrastructures between utilities – including not only systems from different vendors and for different purposes, but also similar systems with different roles and features. Some utilities operate all the systems themselves, while others have service providers that take care of defined sectors such as smart metering data delivery, for example. The challenge for utilities entering into smart metering is the necessity to build a system infrastructure which includes existing systems and new system layers that support their needs in an optimal way. Questions like “Do the data format and validation capabilities between the systems match?“, “Which systems need network status or load profile information?”, “Where are the smart metering installations managed?” or “How and where do we manage the smart meter infrastructure?” need answers. Namely systems with the same purpose, especially Meter Data Management (MDM) systems, have very diverse capabilities to manage various data sources, which complicates the evaluation between different system combinations. The key to success for utilities is to take a look at the big picture, especially the role and purpose of each system, in order to find the best components to support their processes and to complete the IT infrastructure accordingly. Landis+Gyr has delivered more than 300 AMM systems in Europe alone and developed the Gridstream HES system for utilities that need to connect their processes and systems with smart meters. “The challenge for utilities entering into smart metering is the necessity to build a system infrastructure which includes existing systems and new system layers that support their needs in an optimal way." How to hook up a new AMM system to your IT infrastructure 4 White Paper Gridstream HES has four main purposes which serve utilities’ core processes and other systems: 1. Support for automated metering data processes by managing automated smart metering data collection and delivering it to other IT systems. Billing, load profile, event and power quality data are typical examples of forwarded data. 2.Support processes that need to access on time data from the smart meters on demand. Gridstream HES provides quick access from other systems to the on time meter data, typical examples being: R esponding to end-customer claims or questions related to power quality or billing during a phone call S upporting on-site work related to network problems, such as locating power outages 3. Support processes requiring instant notification of critical events detected by smart meters, such as power outages or tampering, by delivering meter triggered alarms to the system where operators can immediately take action to eliminate the problem. 4. Provide applications to manage the smart metering infrastructure Manage meter configurations, settings and firmware Manage automated meter reading processes M onitor smart metering communications and data delivery to other systems With these features, Gridstream HES provides a lean system layer, giving utilities the flexibility to build their core processes and to integrate them with other IT systems while ensuring efficient meter park lifetime management. Furthermore, the system can be integrated into utilities’ internal operations or can be sourced out to service providers. 3. How to integrate easily? After the roles and purposes of the utility’s IT systems have been defined, they need to be integrated. Typically, integration processes between systems are implemented by tailoring the integration layers in each system involved. The challenge of this approach lies in the complexity of How to hook up a new AMM system to your IT infrastructure carefully defining the integration interfaces and responsibilities of each vendor and possible 3rd party integrators. It is likely that in the specification phase not all related details are taken into account and when problems occur it can be challenging to find the responsible party to finalize the integration. Gridstream HES provides an easier solution to manage this challenge. Landis+Gyr has actively participated in the standardization work groups of the IEC 61968-9 standard for smart data integration to DNO systems (see Figure 1). This standard enables any system vendor to use the common language to engineer the integration enabling system component extension or replacements while eliminating the utility’s risk of being dependent on individual vendors. Gridstream HES’ ease of integration doesn’t only rest on the open integration process. The fixed integration layer with ready-made setup, documentation and testing also facilitates the integration process. In this approach, the other systems are integrated to Gridstream HES by modifying the integration layer in their system according to the .xml schemas provided by Landis+Gyr. This integration layer, designed for smart metering data delivery according the IEC standard, is already specified and tested. Integrators save effort and time, since they do not need to specify the layers or the roles of both systems and receive a mature interface they can work with. From a utility’s perspective, this approach is significantly easier to manage: the Gridstream HES system includes a standard interface and the integration service is retained from only one IT system vendor or integrator. These parties receive the integration documentation, technical support and access to the Gridstream HES integration testing system from Landis+Gyr. This also makes it easier for them to offer their integration solution to the utility. In addition, Gridstream HES supports interoperability of smart metering devices by supporting Landis+Gyr’s smart metering portfolio and 3rd party meters based on the IDIS standard. Furthermore, the system features an architecture that enables efficient connection of 3rd party meters to utility processes via Gridstream HES. 5 White Paper Figure 1: IEC 61968-9 coverage (source: “Application Integration at Electric Utilities - System Interfaces for Distribution Management - Part 9: Interfaces for Meter Reading and Control”, IEC 61968-9, Final Draft International Standard, 2012) Area of Direct Impact using IEC 61968-9 (and within scope of 61968) Area Causally/Indirectly Impacted by or impacting IEC 61968-9 Customer Electric Utility Enterprise Applications Standard or Proprietary Communication Infrastructures Enterprise Integration Infrastructure (e.g. ESB, SOA, …) Customer (e.g. CIS, WMS, GIS, MDMS) Operations Applications (e.g. OMS, DMS) IEC 61968-9 Messages (or more broadly any 61968 messge) Meter Head End Systems Messages defined by IEC 61968-9 are based upon IEC CIM and conveyed using a variety of integration technologies and IEC 61968-100 Mappings, translations and/orforwardiing as needed Meter or Gateway PAN PAN Device Customer Messages defined by relevant standards or vendors. May use a wide variety of communication technologies TC57 WG21 Gateway Mapping, translations and/or forwarding as needed PAN Device PAN Device PAN PAN Device PAN Device Messages defined by PAN/HAN specifications 4. Proven platform for optimizing system HW performance and costs When building AMM systems, utility IT managers need to keep in mind more than just the performance of the system; they also need to keep an eye on the lifetime costs incurred by hardware and software licenses. This chapter introduces the results of testing the Gridstream HES with up to 3.000.000 metering points using a variety of simulated data sets. It also considers how the results can be used to optimize the lifetime costs of the system with different scales of AMM systems. 4.1 Test setup description We built the test system setup in cooperation with IBM. IBM provided the recommended hardware setup for the tests and helped to analyze the results from a hardware configuration point of view. Gridstream HES enables flexible system architectures from centralized (all processes How to hook up a new AMM system to your IT infrastructure are run with one server) to decentralized servers (processes are divided between several server computers) or a combination of the two. In the test system, the three key components were run on separate servers (see picture below): 1. Communication server 2. Database server 3. IEC integration server The performance of the system can be increased horizontally by just adding parallel servers for any of these layers. Gridstream HES integration servers use queue based communication to enable connectivity with just one endpoint for integrated systems. This means that adding servers to Gridstream HES does not require changes in other integrated systems. 6 White Paper An additional server was used to simulate data delivery from a point-to-multipoint smart meter park, with a maximum capacity of up to three million metering points with three different data sets (see also Table 1): During the tests, several performance indicators were monitored: data throughput time of each system component from simulator to the integration layer, disk drive performance and CPU utilization. Profile 1 include 3 day registers and 1 hourly register Profile 2 include 2 day registers and 2 hourly registers Profile 3 include 1 day register and 3 hourly registers Figure 2: system setup including simulation and Gridstream HES configuration Head End System Simulator • Windows 2008 Server R2 SP1 (64-Bit) • Oracle 11g DB 64-bit • Oracle 10g IAS 64-bit • Java SE 1.6 64-bit • Model: X 3550 M4 • CPU 2*Intel Xeon ES-2690 2.9GHz (32 core) • Memory 128 GB • SSD drives: C: 146 GB and D: 438 GB Push simulated data to corresponding meter in HES DB Communication Server • Windows 2008 Server R2 SP1 (64-Bit) • Oracle 11g DB 64-bit • Java SE 1.6 64-bit • Model: X 3550 M4 • CPU 2*Intel Xeon ES-2690 2.9GHz (32 core) • Memory 128 GB • SSD drives: C: 146 GB and D: 438 GB Send meter and register values to JMS Queue for 3rd party MDM service HES DB Server IEC Server • Windows 2008 Server R2 SP1 (64-Bit) • Oracle 11g DB 64-bit • Java SE 1.6 64-bit • Model: X 3550 M4 • CPU 2*Intel Xeon ES-2690 2.9GHz (32 core) • Memory 128 GB • SAN disk system V7000 • C: 270 GB, D: 1,68 TB, F: 1,68 TB, G: 1,68 TB and H: 1,68 TB • Apache ActiveMQ Message Broker 2.0 • Windows 2008 Server R2 SP1 (64-Bit) • Java SE 1.6 32-bit • Model: X 3550 M4 • CPU 2*Intel Xeon ES-2690 2.9GHz (32 core) • Memory 128 GB • SSD drives: C: 146 GB and D: 438 GB Utility systems Simulated data delivery for a mass of meters, with different data collection profiles (table 1) Table 1: Data set profiles used in the performance tests Table 1: Data set profiles used in the performance tests Day registers Hour registers Register values per meter / day Register values for 1.000.000 meters / day Register values for 2.000.000 meters / day Register values for 3.000.000 meters / day How to hook up a new AMM system to your IT infrastructure Profile 1 A+, A-, R+ R27 27.000.000 54.000.000 81.000.000 Profile 2 A+, AR+, R50 50.000.000 100.000.000 150.000.000 Profile 3 A+ A-, R+, R73 73.000.000 146.000.000 219.000.000 7 White Paper 4.2 Test results description The data delivery performance of the three server test system was analyzed based on the data throughput time of each server and for the whole system, with each test profile and with data for one, two and three million metering points. The results are presented in the figure below, which shows that each component provides a very linear performance between the data delivery of one, two and three million metering points. All processes were up and running two minutes after starting the simulation, which ensures efficient usage of all system resources. Server CPU utilization was almost flat throughout the tests and the CPU load level was below 50% with Profile 1, and between 50% and 57% for Profiles 2 and 3. The SSD disk operations level was a few thousands of input/output operations per second (IOPS) and frequently reached 6000 IOPS, which is 30-40 times higher than what can be achieved with 15.000 rpm SATA or SAS HDD drives. The disk drive load was higher with Profile 1 than with Profile 2 and 3 tests. Table 2: Total data test system performance Table 2: test system performance (hh:mm:ss) 1.000.000 meters 2.000.000 meters 3.000.000 meters 1:49:00 2:52:00 5:57:00 3:45:00 5:47:00 13:00:00 6:37:00 11:55:00 18:45:00 Total time (Profile 1) Total time (Profile 2) Total time (Profile 3) Graph 1: Data throughput times for each test system component 18:45:00 Total time (Profile 3) 11:55:00 Total time (Profile 2) 06:37:00 Total time (Profile 1) 18:36:00 IEC integration server (Profile 3) 11:31:00 IEC integration server (Profile 2) 06:36:00 IEC integration server (Profile 1) HES DB server (Profile 3) 16:48:00 11:14:00 HES DB server (Profile 2) 05:58:00 HES DB server (Profile 1) 10:48:00 Comms server (Profile 3) 06:46:00 Comms server (Profile 2) Comms server (Profile 1) 04:50:00 Simulator (Profile 3) 04:54:00 Simulator (Profile 2) 04:57:00 03:53:00 Simulator (Profile 1) 00:00:00 3.000.000 meters 02:24:00 04:48:00 2.000.000 meters How to hook up a new AMM system to your IT infrastructure 07:12:00 09:36:00 12:00:00 14:24:00 16:48:00 19:12:00 1.000.000 meters 8 White Paper 4.3 Conclusions drawn from the test From a hardware setup point of view, using SSD drives is the most important factor to improve system performance when the system reads out 50.000 or more metering points. With only three servers, the Gridstream HES test system demonstrated the ability to handle the test profile data sets for up to three millions metering points well within a day, and delivered the first test profile data (three day registers and one hourly register) well within a typical eight hour data SLA. This ensures that the smart metering data is ready for the utility’s processes and operations for the first office hour on a daily basis. Gridstream HES’ substantial data delivery capacity with a decentralized architecture allows us to optimize the system for the diverse needs of utilities. As a result, users can implement various strategies to optimize the system cost, reliability and performance. Focus on performance: A decentralized architecture with efficient system processes enables horizontal scaling, so that Gridstream HES can scale up to meet the extreme performance requirements in different smart metering environments. This is achieved by adding hardware capacity to the most loaded component, for example the communication server. This enables system optimization for various smart metering setups, from pointto-point only to point-to-multipoint systems, which pose different challenges to the communication layer. Based on the tests, some examples of recommended hardware configurations are listed for systems that need to manage Profile 1, 2 and 3 data within eight hours. The IBM® systems used in the test demonstrated reliability, uptime, performance and I/O flexibility. With more computing power per watt, Intel Xeon E5-2600 series processors, advanced memory support, and greater disk capacity, the servers delivered balanced performance and density. The technology used in the storage infrastructure allowed a high storage IO performance and contributed significantly to the test results. IBM Storwize V7000 provides a centralized, integrated platform for optimizing and managing storage resources, driving simplicity of management, reduced cost, flexibly scalable capacity, performance and availability. For more information on IBM infrastructure solutions, please visit www.ibm.com/smarter-computing Table 3: Examples of IBM hardware setup costs for different system sizes targeting at 8 hour SLA with Profile 3 HES server Communication server IEC server Total € Cost MP/€ <50 000 MP HES 8334,00 8453,00 2523,00 19310,00 0,39 <100 000 MP HES 11409,00 8453,00 2523,00 22385,00 0,22 <250 000 MP HES 17224,00 8453,00 2523,00 28200,00 0,11 <500 000 MP HES 23948,00 8453,00 2523,00 34924,00 0,07 <1 000 000 MP HES 33681,00 8453,00 2523,00 44657,00 0,04 <2 000 000 MP HES 61543,00 17086,00 2523,00 81152,00 0,04 <3 000 000 MP HES 89405,00 17086,00 2523,00 109014,00 0,04 (non performance/cost efficiency) How to hook up a new AMM system to your IT infrastructure 9 White Paper Focus on cost: The decentralized architecture allows to set up the system in a way that ensures an optimal cost structure, taking into account direct hardware costs, indirect hardware costs (server room costs such as space and energy consumption expenses) and direct software costs (database and operation system licenses). System performance and architecture allows selecting servers with appropriate number of CPUs allowing the use of affordable Oracle Standard Edition One or Standard Edition database licenses. environment costs vary between 7cnt/per meter to 1.5 cnt/ per meter depending on target performance (Profile 1-3) and system size (500k MPs – 3M MPs). The costs per meter decrease equally in all test profiles when the system size grows from 500.000 to 1.000.000 metering points. Beyond this size, more data requires higher performance, and the costs even out for Profile 3 at one million metering points, and for Profile 2 at two million metering points. Based on the tests we present the system IT costs (server HW, database and OS licenses) for different system sizes and performance targets in Figure 3. The direct server Graph 2: System IT environment (server HW, database and OS licenses) costs per metering point with 8 hour SLA 0,40 0,35 0,30 €/MP 0,25 0,20 0,15 0,10 0,05 Cost MP/€ <50.000 MP <100.000 MP <250.000 MP <500.000 MP <1.000.000 MP <2.000.000 MP <3.000.000 MP Profile 1 0,39 0,22 0,11 0,07 0,04 0,02 0,01 Profile 2 0,39 0,22 0,11 0,07 0,04 0,02 0,02 Profile 3 0,39 0,22 0,11 0,07 0,04 0,04 0,04 How to hook up a new AMM system to your IT infrastructure 10 White Paper Focus on reliability and maintainability: Utilities can optimize system reliability by ensuring quick recovery in case of a system hardware component failure. System maintainability and related costs can be optimized by system configuration design that is based on hardware components which are readily available and have low stand-by costs. To give an example: for large systems, a communication server layer can be built by using several low cost servers (instead of a single high performance unit) commonly used elsewhere in the utility’s IT environment, which enables keeping a pre-configured spare unit on the shelf for quick recovery. Based on research like the tests presented above, Landis+Gyr can help utilities of any size to find the best system setup to support their IT strategy and needs. The system is hardware brand independent, which allows us to provide recommendations based on the utility’s preferences. Author: Jarno Suomela, Product Manager, Landis+Gyr In February 2013, Landis+Gyr was among the winners of the 2013 IBM Beacon Award for Best Industry Solution for Energy & Utilities. Every year, this honor is awarded to recognize a select number of IBM Business Partners who deliver innovative solutions and set the standards for business excellence, ingenuity, and an exceptional client experience. About Landis+Gyr A trusted name in energy management solutions, Landis+Gyr operates in 30 countries across five continents. Landis+Gyr ranks as the worldwide leader in electricity metering with a preeminent position in Advanced Metering Management. Its meters and solutions empower utilities and end-customers to improve their energy efficiency, reduce their energy costs and contribute to a sustainable use of resources. With a proven track record for more than a century, it’s Landis+Gyr’s primary goal to help utilities manage energy better. For more information on how you can manage energy better please contact: Landis+Gyr AG Theilerstrasse 1 6301 Zug Switzerland Tel: +41 (0)41 935 6000 Fax: +41 (0)41 935 6501 Email: info@landisgyr.com Website: www.landisgyr.com/europe How to hook up a new AMM system to your IT infrastructure 11