Selecting a Building Management System (BMS) for Sites with a Data Center or IT Room

Overview

The paper argues that a BMS serving a site with a data center or IT room must go well beyond comfort-oriented commercial BMS functions. Required capabilities are automatic physical response to cooling failures, end-to-end visibility from utility and mechanical plant to servers, awareness of redundancy and dependencies, capacity tracking, contextual alarming with attached procedures, data sharing with EPMS and DCIM, and strong cyber security. When a cooling unit fails, data center temperatures rise rapidly in the first 60 seconds, so the BMS must start redundant cooling automatically.

The operating-hours gap matters too: a commercial building occupied 7 am-6 pm on weekdays runs about 2,860 hours a year versus 8,760 for a data center, so efficiency changes yield roughly 3x the savings of comfort cooling. Human error is attributed with 70% of data center outages (Uptime Institute AIR database), split into active and passive errors. Verizon DBIR 2016 is cited: 63% of confirmed breaches used weak, default or stolen passwords, and 70% of insider-misuse breaches took months or years to discover.

The paper positions the BMS alongside EPMS for electrical monitoring and DCIM for the white space, connected over open protocols such as BACnet IP, MS/TP, Modbus IP and RTU on Ethernet TCP/IP, plus web-services APIs. The recommended split is DCIM for white-space applications, BMS for building and mechanical, and EPMS for electrical, because each UI is optimised for one domain. It closes with three implementation pitfalls (wrong tool, weak processes, lack of ownership) and four critical operator processes: inventory and asset management, system configuration, alarm integration and stakeholder reporting. Facilities and IT should jointly witness commissioning tests and agree pass/fail before go-live.

Key points

1. After a cooling unit failure temperatures rise rapidly in the first 60 seconds; the BMS must start redundant cooling automatically.
2. A commercial building runs about 2,860 h/year versus 8,760 for a data center; efficiency gains are roughly 3x.
3. Uptime Institute AIR: 70% of outages attributed to human error.
4. Verizon DBIR 2016: 63% of breaches used weak, default or stolen passwords; 70% of insider-misuse breaches took months or years to discover.
5. Open protocols: BACnet IP, MS/TP, Modbus IP and RTU on Ethernet TCP/IP, plus web-services APIs.
6. Split: DCIM for white space, BMS for building/mechanical, EPMS for electrical.
7. Four critical operator processes: inventory, configuration, alarm integration, stakeholder reporting.
8. Facilities and IT should jointly witness commissioning tests.

Using it in design

• In the monitoring design module let users scope BMS, EPMS and DCIM separately and assign each monitored asset (chillers, CRAHs, UPS, PDUs, rack PDUs) to one system with a protocol (BACnet IP/MS-TP, Modbus TCP/RTU, SNMP).
• Require every redundant cooling unit to have an automatic failover control point.
• Warn when the power and cooling chain from utility to rack has unmonitored nodes; require the BMS network to be segmented from the corporate or building network.
• Generate points lists and alarm matrices with severity, recipient role and linked MOP/EOP; add a joint Facilities/IT witness step to the commissioning checklist.

Who should read it

Designers and operations managers scoping BMS, EPMS and DCIM for sites with a data center.

Related standards

• ISO 16484-5 (BACnet)
• IEC 62443
• EN 50600-2-6
• ANSI/TIA-942

Publication details

Authors

Bryan Anderson; Patrick Donovan

Publisher

Schneider Electric

Year

2016

Document number

WP 233

Document language

English

Source type

Manufacturer

Access

Free to read

Licence

Copyright retained; citation only — the publisher’s terms apply to the original document

Citation