A UPS system must have communications capability to provide full power protection for its critical load. Communications-enabled UPSs can warn their protected devices of an extended power outage, allowing them to shut down safely during the UPS system’s battery autonomy. Apart from this basic and essential requirement, UPS communications allow data centre managers to run entire systems with multiple UPS modules safely and efficiently- even if the network is large or extends across multiple sites. In this article, Stuart Cockburn, senior field sales engineer at Uninterruptible Power Supplies, a Kohler company looks at the different hardware and software communications components available to UPS users. He then discusses how these can be integrated into a solution that best meets an individual data centre’s requirements
An uninterruptible power supply (UPS) system that does not communicate with its critical load or appropriate monitoring station cannot provide a full power protection function; without the benefit of a warning, the critical load will crash if a power outage ever exceeds the UPS battery’s autonomy. The UPS will have merely delayed rather than prevented the inevitable damage that ensues.
Accordingly, most users regard a UPS’s communications capability as one of its essential features. An early warning from a UPS of an impending serious problem allows operators to benefit properly from the battery’s autonomy. They have time to either shut the critical load down safely or, if continued availability of the IT resource is critical, start up a standby generator.
All modern UPS systems therefore have a number of communication options for signalling their operational status and activity to remote monitoring stations, building management systems and the critical load. These range from simple volt-free contacts to sophisticated systems that communicate over serial links or across site networks and the Internet. Users can choose the options most appropriate to the size, complexity and criticality of their data processing resource. The remainder of the article helps managers in making these choices by looking more closely at the communications solutions available to them.
Volt-free contacts – a simple solution
Alarms and status are usually signalled on the UPS’s front panel display. These signals should also be made available as simple volt-free contacts which can be connected to alarm panels, monitoring systems, remote status panels (RSPs) or the site’s building management system (BMS). They provide basic True/Not True information for easy display where required. Their role is not limited to passive signalling, however; if the UPS detects an unacceptably extended power outage, one of its contacts can initiate an orderly shutdown sequence for the protected PC or network. Almost all Windows, Mac, Linux and UNIX variants support suitable software, which can be supplied by the system, network or UPS vendor.
While shutting down the computer system is essential at the onset of an extended power failure, it is equally important that short-term electrical disturbances are prevented from causing such shutdowns. Accordingly, timers are used with the contacts to ‘filter out’ power events that do not threaten UPS battery autonomy. Network shutdowns can be handled on a Master/Slave basis. A computer network server can be assigned as ‘master’; this can relay UPS alarms to other network devices or ‘slaves’ configured to respond appropriately.
Serial communications and Modbus
Many UPS systems have serial RS-232 connections which can transmit more detailed information than the volt-free contacts’ True/Not True status indication. A remote terminal could, for example store and display voltage, frequency, current and power measurements taken from the inverter and bypass switch outputs, battery condition and status, and statistics related to mains failures and UPS operation.
The exact characteristics of the RS-232 link vary among UPS manufacturers. This is because there is no European standard for UPS RS-232 protocol, so users must obtain manufacturer-specific shutdown software from the UPS vendor. Most manufacturers endeavour to maximise the serial link’s value with varying additional features. In normal practice, critical alarms are continuously monitored while UPS operational status is requested manually by a system administrator or supervisor.
In any case, users will need monitoring software to display the information received via a serial link; such software is available for most operating systems. A UPS monitoring screen typically shows a graphical display of UPS status, voltage, current, load, battery voltage, power frequency and other parameters, as shown in Fig.1.
Fig. 1: Typical UPS monitoring screen
Other activities can also be set up – these include configurable responses to certain alarms, message broadcasts, scheduled diagnostic checks and data logging. If communication distances are longer, RS-485 or RS-422 interfaces can be used instead of RS-232.
An applications layer serial protocol known as Modbus or JBus is available for installations where individual UPS modules must be addressable with precision. Modbus systems of up to 240 nodes can be implemented across either serial RS-485 links or an IP network. Each node can be a data processing device or UPS module and each has a unique address. In most implementations, a single master is used to address all the other devices on the bus, configured as slaves.
Each UPS slave can store UPS input current, voltage and other information; the master device can interrogate the slaves, obtain this information and feed it to a central UPS management station. The master can also address and send commands to individual UPS modules. Alternatively it can issue ‘broadcast’ commands which are actioned by every module on the bus. These commands are typically used to transfer all of a UPS system’s modules to bypass or initiate an emergency power down.
Larger systems and network communications
As UPSs are called upon to support increasingly larger-scale computer installations, they encounter greater network complexity and often wider distribution of data processing elements. This means that using network communications for the UPSs themselves becomes the most convenient approach to management. For example a managed and co-ordinated enterprise-wide response is necessary for handling power fail situations safely and effectively.
During a power failure, monitoring software will gauge each UPS’s remaining battery time, and initiates a shutdown of the UPSs’ protected computers when necessary. During this process, a common UPS supported network environment is needed to integrate the various monitoring solutions used by different manufacturers’ UPSs. In these circumstances, Remote Control Command (RCCMD) is the most frequently used client solution.
Under RCCMD, the monitoring software can send out an RCCMD shutdown signal at a predetermined level of remaining battery run-time. Any remote PC or server with an installed RCCMD client licence will shut down gracefully in response to the signal. The IT system can be configured so that less-essential units are shut down first, allowing the battery energy and back-up time to be conserved for the more critical items.
The challenge for data centre administrators intensifies if their Wide Area Network (WAN) system is spread over a number of sites, some without a network manager. A problem occurring on an unmanned site could cause irreparable damage to local hardware and software if no staff are available to contain it. The most common solution is to use Simple Network Management Protocol (SNMP) as this allows a network manager to monitor and control all network devices from a central location.
SNMP is a standard protocol that was created as a solution for wide area network management. As part of the TCP/IP suite, it allows all network devices to transmit management variables across enterprise-wide networks. Adding SNMP to a UPS makes it intelligent, allowing it to perform many tasks including logging events, continuously monitoring power quality and reporting on battery health. It can also perform and report on self-diagnostics. Additionally, intelligent UPSs can activate or deactivate individual devices. Areas can be isolated for security, loads can be powered down to cut energy use, and sectors can be managed for redundancy.
As well as providing real-time information for immediate analysis and action, SNMP allows data centre managers to spot trends and potential problems such as battery issues or persistent power events. This allows the problems to be forestalled with corrective action. The SNMP environment also allows managers to run their network on an enterprise-wide basis, rather than simply controlling individual UPS units. Large numbers of UPSs can be balanced against one another as an integrated power protection resource, and managed from a central network console. In parallel redundant UPS systems, for example, a system shutdown can be avoided if just a single redundant UPS module fails.
To access to the SNMP environment, a UPS system must become a network peripheral device. It does so by using an SNMP adaptor. These adaptors usually interface the UPS serial communications port to the network. SNMP adaptors that accept inputs from the UPS’s volt-free contacts are also available. Any adaptor chosen must be used with network operating system compatible software.
The choice of SNMP adaptor also affects the level of monitoring, control and shutdown facilities available. Accordingly it is important to understand how well these meet the requirements of a target installation before making a final selection. As well as remote control, rebooting and safe shutdown of UPS-protected devices, other functions may be desirable. These include an ability to deliver real-time email, mobile phone or SMS alarm notifications, data encryption and authentication, automatic delivery of local-language web pages, and activity and alarm logging.
In addition to enterprise network connectivity, some modern UPS systems can signal fault indications and maintenance requirements to remote monitoring points via SMS, mobile phone or the Internet. Once alerted, a service engineer can perform basic, remote UPS diagnostics. If appropriate, he can attend the UPS on site, fully-stocked with the UPS components needed to effect the repair.