Chassis management for VPX systems

2Systems built to VITA standards, such as VME, have not traditionally needed the high reliability of many of the telco systems, and have relied upon front panel LEDs to indicate fan failure, voltage monitoring, and thermal issues with the boards and chassis. While VME has long been the preferred architecture of the military and aerospace industry, as higher speeds and increased reliability became more critical to system operation, it was apparent that these older architectures just wouldn’t cut it. VPX, or VITA 46.0, was VITA’s answer to implement modern high-speed interconnects, real-time monitoring, and remote mitigation of certain system parameters in harsh environmental applications.

However, due to the large number of possible configurations, the interoperability of these VPX-based systems was questionable. VITA introduced OpenVPX, which addressed many of the interoperability and configuration concerns.

The next iteration was VITA 46.11 “System Management on VPX” specification, released in November 2013 as a draft standard for trial use. VITA 46.11 leverages the shelf management functionality that has been used for years in the telco industry to add reliability and robustness to typical VPX applications.

Chassis management

VITA 46.11 brings a new level of chassis management to VPX as an advanced monitoring tool versus using LEDs as indicators of problems. Prior to this specification, the various VITA systems had very limited management or monitoring capabilities unless custom software was written to interface with the boards. For example, the only real monitoring in VME systems was from the utility bus signals:

  • System Clock (SYSCLOCK)
  • AC Fail (ACFAIL*)
  • System Reset (SYSRESET*)
  • System Failure (SYSFAIL*)
  • Serial Bus A (SERA)
  • Serial Bus B (SERB)

While there are various timing requirements for these signals that make sure the boards come up when all the boards are ready, the only real oversight traditionally done on these systems was to monitor the backplane voltages, and sometimes the fan speed (or a fan fail signal) or the chassis temperature. Without elaborate middleware there was no central location where a developer could consistently monitor critical parameters, such as board temperature and health, backplane voltages, chassis temperature, or fan speed.

The new VPX chassis manager enables remote access to any board and chassis that supports VITA 46.11. Since the VPX shelf manager is based on PICMG’s ATCA version of shelf management, a brief overview will provide some needed background.

CompactPCI shelf manager

PICMG started using shelf management in the 2000s, when the PICMG 2.9 CompactPCI Shelf Management specification was released. At this point, the shelf management primarily consisted of maintaining an inventory, logging events, and monitoring sensors. For CompactPCI, there are only 18 IPMI commands that need to be supported.

IPMI, or intelligent platform management interface, is a standardized computer system interface or protocol that communicates across an I2C bus connected to each board, the shelf manager, and any other intelligent FRUs (field replaceable units). CompactPCI provided for one IPMB (intelligent platform management bus) for communication.

ATCA shelf manager

The shelf management function was expanded with the ATCA (PICMG 3.0) specification, providing extensive management capabilities. These include low-level hardware management service, high-speed management services based on TCP/IP protocol, and in-band application management. Because it was geared towards the telecommunications industry, ATCA was designed for redundancy using two IPMBs, which could be bussed or radial.

While low-level management consists mainly of monitoring power, cooling, and interconnect resources, the ATCA shelf manager watches over the basic health of the system, reports anomalies, and takes corrective action when necessary. It monitors, controls, and ensures the proper operations of the ATCA boards and any other chassis components. The shelf manager can also retrieve inventory information as well as receive event reports and failure notifications. If necessary, it can perform basic recovery operations including resetting or power-cycling boards.

VITA 46.11 (VPX shelf management)

By leveraging the inherent capabilities of the PICMG ATCA shelf manager, VITA is reducing a great amount of the preliminary development work for the implementation of the VITA 46.11 chassis manager.

One of the major differences between the PICMG 3.0 shelf manager and the VITA 46.11 chassis manager is that PICMG 3.0 requires all boards to have IPMCs (IPMI controllers), however, IPMCs are optional for VITA 46.11 boards.

The standard VPX system does not require system management to function and hot swapping is not required, which precludes the ability of the VPX chassis manager from providing power management. Since the chassis manager does not provide power management and not all boards will have IPMCs, electronic keying (E-Keying) cannot be used for the VPX systems.

However, the VITA 46.11 specification does state that there may be some level of E-Keying in the future, and the existing VPX specification does provide for physical keying to prevent plugging incompatible boards into the wrong backplane slot.

The primary functions of the VITA 46.11 chassis manager are: inventory management; sensor management; system configuration; recovery; and diagnostic management. The three VITA 46.11 chassis manager layers are: IPMC; chassis manager; and system manager. These management layers are hierarchical in nature, where the IPMC (board level management) communicates with the ChMC (chassis manager), which, in turn, reports to the system manager. The system management layer monitors multiple chassis.

IPMC (IPMI controller)

The lower logical layer of management would be the IPMCs, which are required on all intelligent FRUs, such as front loading VPX plug-in modules, fan trays, power supplies, etc. An IPMI controller for boards or intelligent FRUs is used to monitor: health of the board or FRU; voltages; temperature; device ID; serial numbers; part numbers; and software versions. The SDR repository will provide a full list of all the sensors on a particular board or FRU.

Figure 1 shows a block diagram of an IPMC where the Freescale processor has connections to the IPMB and the payload through both I2C and analog I/O, providing information on most of the sensors. This IPMC also has links to the DC/DC input power enabling power management to the hot swap controller, allowing for hot swapability as well as links to LEDs on the front panel. The IPMC shown will support cPCI, ATCA, AXIe, and VITA 46.11 applications.

Figure 1: The IPMC monitors several parameters simultaneously for better system management.
(Click graphic to zoom)

Chassis management controller (ChMC)

Next up is the chassis management controller (ChMC), which could be implemented in several ways. It could be implemented on one of the front-loading VPX plug-in modules, as a mezzanine board that plugs onto the backplane, or as a standalone board.

The lead image of this article shows a VITA 46.11 chassis manager next to a 3U x 160 mm carrier card. In this implementation, the chassis manager could be mounted to the carrier card allowing for a pluggable interface.

The advantages of any VPX chassis management controller are numerous:

  • Redundancy. While not a requirement for a VPX chassis manager, this would make the system more robust.
  • Simplified power management. VITA 46.11 does not require E-Keying or power management, but they may be useful in the future. And since ATCA requires them, most VPX chassis managers will already support this.
  • Cooling control. Although several methods could be used, a typical method is to monitor board and chassis temperatures, then adjust the fan speed to maintain the predetermined range.
  • Inventory management. The chassis manager maintains a full list of all intelligent FRUs and boards as well as any other components that support VITA 46.11 (mandatory support of VITA 46.11 in VPX boards is not yet a requirement).
  • Sensor management. A list of all sensors connected to each intelligent FRU in the system, along with any threshold or limits, is maintained via the chassis manager.
  • Sensor event log. Although the actual log size will vary among chassis managers, it provides a history of all events such as an over temperature condition or an under voltage condition and typically will begin overwriting when the log is full, starting with the oldest event.
  • Diagnostics and recovery. The specific VPX boards in a system, and their compatibility with VITA 46.11, will determine the level at which the chassis manager can diagnose and respond to system events.

System manager

At the top of the logical management layer is the system manager, which oversees multiple chassis and will communicate with multiple chassis managers. The system manager can be middleware, a SNMP MIB browser, custom software, RMCP, or something as simple as the ability to Telnet into a system.

The system manager is how the outside world communicates to the VITA systems via the chassis manager. Since the chassis manager supports several external interfaces via RS232 and Ethernet interfaces, many also have a built-in GUI that could be used for a lower level of system management.

The system manager provides the ability for users to remotely access the VITA 46.11 chassis to manage board inventory, and monitor the health of the boards as well as the health of the chassis and all manageable FRUs in the system.

While a VITA 46.11 chassis will typically have intelligent FRUs, such as fans and power supplies, the challenge for board manufacturers is to implement IPMCs on their boards, allowing for remote monitoring. This will further enable boards and IPMCs that support VITA 46.11 to be easily inventoried and that are monitored for health and, in some cases, the boards may be remotely reset, if necessary.

Advanced system management

As more intelligent monitoring functions are implemented into VPX applications, existing shelf management capabilities will continue to increase the effectiveness, value, and reliability of these embedded systems well into the future.

Gary Hanson is a Sr. Systems Engineer for Elma Electronic.

Elma Electronic 510-656-3400