Suvish Viswanathan is the senior analyst, unified IT at ManageEngine, a division of Zoho Corp. You can reach him on LinkedIn or follow his tweets at @suvishv. This is a last part of a three-part series.
In the second post in this series, we looked at the evolution of data center asset management and the degree to which it has evolved in parallel with traditional IT management. Ultimately, if you adopt a service-oriented management focus, the goal of both management efforts is the same — to enable the optimal delivery of a service to the end user. That said, the IT and facilities management worlds should not need to operate in parallel. Instead, they need to operate as one, in a truly integrated manner.
That’s what DCIM — data center infrastructure management — should be all about.
Now, before you go saying “Ah, DCIM — it is rubbish” (as, in fact, a journalist said to me just the other day), let me distinguish what I’m talking about from the DCIM that everyone else is talking about (which, I agree, is rubbish).
Unfortunately, DCIM has become one of those buzzwords in the marketplace that has no standard definition. I recently saw one article which mentioned that more than 80 vendors claim to offer DCIM solutions. The problem with that is that most of them don’t. They may offer an IT or facilities management product that facilitates the management of one part of the data center infrastructure, but that’s a far cry from the kind of integrated DCIM solution that today’s fast-paced business needs.
The Shape of a Truly Integrated DCIM Solution
Data center management will never be performed efficiently if the IT infrastructure and facilities infrastructure are managed separately. Can you imagine your blade servers running in a room cooled to only 90°F? Should you really feel comfortable about the ongoing availability of your business-critical applications if you don’t know that the diesel tank fueling your back-up generator is only 10 percent full? Is it really possible to ensure the security of your infrastructure and the critical data it processes without a proper sensing mechanism in place?
When viewed through the lens of service delivery, all the assets in your data center are connected, and your ability to monitor and manage them needs to be equally as interconnected. A true DCIM must be able to do the following:
Collect Data. The data center is full of data collection nodes: IT systems collecting performance data in real time from servers, switches, data storage systems and more — as well as facilities infrastructure systems collecting data about rack temperatures, power consumption, backup generator fuel tank levels and more. These systems rely less and less frequently on an agent-based approach to reporting, so a DCIM solution must be able to collect data using a wide range of common communications protocols — from SNMP, WMI, SSH and the like for IT assets to Modbus, BACnet, LonMark and others for the facilities infrastructure assets.
The data capture features of DCIM need to support more than real-time infrastructure monitoring, too. The DCIM system must be able to reach deep into the broader infrastructure to pull granular data from individual pieces of equipment for planning and forecasting purposes.
Provide analytical support. Ultimately, the point of collecting data is to subject it to analysis and correlation, so a DCIM system needs a powerful analytical component. From a data center management standpoint, the analytical engine can facilitate decisions. These can be programmatic decisions, as when an alert might prompt the automated transfer of virtual machines from one server to another or automatically increase the airflow within a certain set of racks because of a sudden spike in CPU temperatures. Or, they can be strategic decisions taken by a committee, as when planners view DCIM data for environmental trends, application performance patterns or the broader user experience.
Accommodate the operator. A DCIM solution that can monitor and manage a wide range of assets — but only if those assets have been built by the same vendor that built the DCIM solution — is a non-starter. The days of a monolithic, single-vendor infrastructure are long past. In fact, just the opposite is true: The whole notion of the “data center” itself is becoming more and more fluid. If the data center is where an organization runs its mission critical applications and manages the delivery of the user experience, then parts of that data center may be in the cloud. Parts of that data center may reside in physically non-contiguous locations. And decisions about future data center elements may be governed as much by time-to-service delivery as physical location.
An integrated DCIM solution must accommodate a wide range of systems, tools, protocols and standards. It needs to be able to pick up alerts from different assets in the data center and send them to the appropriate authority (via email, SMS or whatever mechanism is preferred by the enterprise). All the elements in the infrastructure need to expose their APIs so that the management tools can understand and interact with them. This would give data center managers the flexibility they need to expand in the ways that will be best for their business (which a vendor lock-in never does).
Control and Automate. Today’s data centers are enormously complex. Some management issues need human oversight; others do not. A truly integrated DCIM solution can help you manage your resources so that issues that do require human intervention are flagged and escalated accordingly. The solution needs to be able to contact the person with the right skills, the right authority and the right access. It needs to be able to alert that person in a manner that is in keeping both with the severity of the issue and the policies and procedures of the organization itself.
For those issues that do not require human intervention, the DCIM must be able to handle them programmatically through various workflow automations. This enables you to focus your (highly intelligent, creative and skilled) human resource on the strategic management tasks that can enhance business productivity, the end-user experience or some other area that matters more to the enterprise.
Manage inventory centrally. Asset management is a major pain point in the data center, but a truly integrated DCIM solution can eliminate this pain through an automated asset discovery engine.
Such an engine would provide capabilities to crawl the data center infrastructure and discover all the devices and services involved — then feed those discoveries into a centralized repository such as a configuration management database (CMDB). Such a database would not be a mere manifest of detected devices, systems and services, though; for this database to be truly useful, it must enable data center managers to understand the relationships between the devices, systems and services. Thus, if a data center manager were planning a project to swap out a row of batteries, for example, the CMDB could let the manager know precisely which servers this row of batteries is backing up as well as precisely which mission-critical applications and services are running on those servers.
The practical impact of any asset change could be readily seen if this kind of DCIM were in place. It’s a hyperconnected world in the data center, which is why we need a truly integrated DCIM tool to handle it.
The Utility of Metrics
Finally, you’ll note that I have not mentioned any of the metrics we usually discuss when talking about data center management. Historically, many people have described data center management in terms of total cost of ownership (TCO), power usage effectiveness (PUE), data center infrastructure efficiency (DCIE) and other metrics. These are important metrics insofar as they can help a data center manager monitor and understand the data center from an environmental perspective. The green IT initiative is important, and failure to monitor with an eye toward the data center’s carbon footprint will have a significantly negative impact on both the company’s tax bill and public image.
However, these metrics provide only a fragmented view of overall data center performance. Data center infrastructure management needs to transcend that fragmented view. The data center is the nerve center of business today, and it needs to be managed with the organization’s service delivery goals in mind. There are human, resource and environmental components that we need to balance and manage effectively. Only by taking an approach that unifies, integrates and consolidates all these elements can we manage the entire data center in a manner consistent with our broader service delivery goals.
Industry Perspectives is a content channel at Data Center Knowledge highlighting thought leadership in the data center arena. See our guidelines and submission process for information on participating. View previously published Industry Perspectives in our Knowledge Library.