Managing the Infrastructure Management Elephant
sking six data center managers to define infrastructure management is a sure way to get six different answers. It is similar to the ancient story of the six blind men and the elephant. When each of the men approached the elephant, he came to the conclusion that the elephant was like a wall, a spear, a snake, a tree, a fan, and a rope. In the end, they were all wrong. Just as an elephant is more than the sum of its parts, a true data center management solution is more than the sum of the various infrastructure management tool sets.
Information is the word that best summarizes the task of infrastructure management. Information about the data center must be complete and accurate, have audit capability, and provide an integration layer.
Data center managers must be able to answer some key questions:
How long does it take to deploy a new server?
Who should be contacted for an accurate status report for a project?
On which servers do any given IT task rely to operate properly?
- Does the data center have the capacity (power, cooling, and space) to add ten new blade servers?
A data center operator who can’t answer these questions probably doesn’t have the tools needed to manage the facility.
Like the six parts of an elephant, infrastructure management tools fall into six general categories:
- Maintenance management
While many of the available tools cross over into multiple infrastructure management categories, it is important to note that no vendor today has a single tool, or even a suite of tools, that meets all these requirements.
Real-time monitoring is almost an elephant in itself because its definition is very dependent upon the perspective of the user of the information. The three primary areas of real-time monitoring are:
Network management system (NMS)
Building management system (BMS)
- Data center monitoring system (DCMS)
Of the three, the first two areas are fairly well understood. In addition the market includes a number of excellent tools. NMS include CA Unicenter, CiscoWorks, HP OpenView, IBM Tivoli, Ipswitch WhatsUp Gold, and Hitachi IT Operations Analyzer are some of the tools found in many data centers. Familiar BMS systems include Honeywell Excel, Johnson Controls Metasys, Siemens Apogee, Andover Continuum, Liebert SiteScan, and Tridium Niagara.
The DCMS is the wildcard. While BMS tools originated as a way to monitor and control equipment in an entire building with data center facilities management as an additional function, DCMS tools were designed from their inception to monitor data center infrastructure equipment and, in some cases, IT equipment as well. Vendors of these crossover tools include both well-known stalwarts in the data center industry as well as a number of new players. Some of the common tools include APC ISX Central, Eaton Foreseer, Modius DCiM, Optimum Path Visual Data Center, Raritan dcTrack, and Geist Environet.
In addition, some hardware manufacturers offer tools to monitor and manage their own hardware products. These products are point solutions that focus only on one hardware platform and, as such, may not integrate well into an overall management solution.
Real-time monitoring systems rely on a number of data collection points and methods including both software and hardware. Most data center infrastructure provides some means for retrieving data and alarms. SNMP, Modbus, OPC, LON, BACnet, and IPMI are the primary communication protocols. Environmental monitoring is an area that is growing rapidly in the data center arena. Some prominent environmental monitoring solutions include APC NetBotz, Geist IT Watchdog, RLE Falcon, Sinetica Hawk-I, and SynapSense LiveImaging. Wireless sensors are becoming more prevalent in the data center with solutions from SynapSense and Arch Rock for monitoring power and environmental conditions.
Asset and change management are important functions in the data center. Management must have an intimate knowledge of the equipment running in the data center. Smaller, less sophisticated organizations typically try to handle these functions using spreadsheets and Visio or AutoCad drawings with coordination handled through email and infrequent change meetings. This is not necessarily bad; it really depends on the size of the data center and, most importantly, on the accuracy of the data. The decision to move to an asset management system is usually driven by the total number of assets, the frequency of moves/adds/changes, the processes involved in keeping the data accurate, and the need to know the relationships between the assets.
While spreadsheets, custom databases, and visual organizers are relatively inexpensive, they usually provide less functionality and require higher management. Asset management software, on the other hand, provides a much higher level of functionality but can be moderately-to-highly expensive.
Cost and time to gather the data the first time are often overlooked when first implementing an asset management system. This is a critical step in ensuring the initial data are accurate. In order to maintain accurate data, it is essential that change control processes be followed strictly.
Some of the prominent products in the asset management arena are Align AssetPoint, AlphaPoint AssetCentral, Aperture VISTA, Avocent MergePoint Infrastructure Explorer, nlyte Software, and Rackwise Data Center Manager.
Radio frequency identification (RFID) technology is being used to automate the tracking of assets in the data center with products such as CTI Asset Tracking System and RFCode Asset Manager.
CFD modeling is a key tool in the initial design of a data center but should also be used to evaluate proposed changes such as increased rack densities or adding, moving, or removing equipment. A data center designed in 2003 may not be configured correctly for IT equipment in use in 2010. CFD modeling can help compare various cooling approaches, provide a view an airflow model of the computer room, and anticipate potential hot spots based on mathematical modeling.
Future Facilities 6SigmaDC, Mentor Graphics FloVENT, Innovative Research TileFlow, and Applied Math Modeling CoolSim are the primary CFD modeling products for data centers.
CFD is not for the faint of heart. The tools are relatively expensive and require some expertise in correctly interpreting the results. Before deciding whether to purchase a CFD modeling tool or to use a consultant trained in CFD modeling, consider how often the organization will need a CFD analysis and whether or not it has staff trained to correctly interpret the results.
While KVM over IP, serial over LAN, console port servers, service processors, and remote control software provide flexibility and out-of-band functionality, these access points also add management complexity. A centralized access management provides centralized monitoring of servers and network hardware, real-time, out-of-band BIOS-level access and control, and simultaneous access by different users from different workstations. Products in this space include Avocent DSView3, Epicenter Centerline, Manage Operations IT & Server Management, Minicom KVM.net II, and Raritan Command Center NOC.
Access Management is not constrained solely to KVM solutions. Most next-generation servers and blade computers have a management or service processor on the motherboard or on a plug-in card. These processors provide a connection to the server’s management processes and can be used to monitor and control the power, provide sensor information (temperature, current, fans, etc.), redirect serial console, and manage the BIOS, as well as to provide full remote control using embedded Virtual KVM over IP. Some of the offerings in the market are:
Intelligent Platform Management Interface (IPMI). HP, IBM, Dell, Sun, and many others
IBM Remote Supervisor Adapter (RSA)
HP integrated LightsOut (iLO)
Sun Advanced Lights Out Manager (ALOM)
- Dell Remote Access Controller (DRAC)
Virtual machines also have their own management consoles. The major management consoles include offerings from VMWare, Citrix, and Microsoft. The focus for access management today is on merged connections that enable a user to view a unified, centralized management console that displays all of the types of access and control regardless if it is physical or virtual.
While sometimes considered a subset of asset management, connectivity management plays a large enough role to warrant its own management category. With connectivity management tools, data center management can track the relationship between assets-both power and network-and trace the connectivity path from beginning to end. This is a valuable source of information for helping to resolve issues that may actually be caused upstream from the device or to prevent issues from occurring downstream from the device. Products in this space include iTracs Physical Layer Manager, Patchmanager PatchManager, and Raritan dcTrack.
Automated connectivity management systems such as the CTI Connectivity Management System can automate tracking of connector changes using RFID tags in the cables and specialized patch panels that act as RFID antennas.
Tracking maintenance of critical data center equipment has been largely neglected in the data center arena. While there are a number of computerized maintenance management software (CMMS) solutions, only PTS Data Center Solutions DCMMS focuses on the data center. It provides the ability to schedule and track preventive maintenance on data center infrastructure, create work orders, and keep track of assets and spare parts required for maintenance. It also provides a number of reports to track costs and system downtime due to maintenance activities.
Where is Infrastructure Management Going?
Infrastructure management tools are constantly changing to fill ever-changing needs. New tools continue to emerge, and current tools are enhanced to provide new functionality. More important, the lines between the functional areas are becoming increasingly blurred as vendors begin to understand that data center managers are looking for a holistic tool to manage the entire infrastructure. Infrastructure management vendors are seeing the need to open up their tools for integration with other product segments, so the data from an asset management system can be used in a CFD model or the output of a real-time monitoring tool might be used to provide information for condition-based maintenance management. The result will be more efficient, more actionable, and more accurate information.