The art and science of leaving the spreadsheet behind
By Todd Goldman, VP of product management at nlyte Software
No data center manager wants to get that call from their CIO asking what the @#**@!! is wrong with some critical system. When the call does come, you want to be able to react quickly, find and fix the offending system and get back to running your data center in a controlled, proactive manner. Of course, that isn’t how things really work today. Normally, you take that call and run to one of tens or hundreds of spreadsheets to see where that one server that is the source of your cascading outage is. Most likely, the spreadsheet is outdated and the culprit has long been moved to an unknown location, sending you off on a hectic pin-in-a-haystack search.
This is how most data centers have been managed (mismanaged?) for the past few decades, but today they’ve become too critical for the status quo to continue. Would your CFO close out your quarter-end or year-end financials using a spreadsheet and the back of an envelope? Of course not! They would use Enterprise Resource Planning software. Does your sales team manage clients and prospects using a spreadsheet? Of course not! They use Customer Resource Management software. If you answered, “Of course not!” to any of these, you should be asking yourself: “Where is the software tool designed to make my life easier?”
Your answer is Data Center Infrastructure Management (DCIM) software. It automates the management, control and predictive intelligence required to optimize the use of power, cooling and space within a data center.
Getting started with DCIM
The first challenge many companies face with DCIM is it will take too long or cost too much. They think DCIM is complicated. And while a fully mature DCIM implementation can be so, it’s not necessarily the case in the early stages.
Like any other software solution, the value of your DCIM implementation is directly related to how disciplined your staff is at using it and how tightly integrated it is with other IT processes and software.
While your end goal might be to reach the upper right quadrant of the DCIM Maturity Model (see FIGURE ONE above) you don’t need to have a fully mature DCIM deployment to get tremendous value out of DCIM. In fact, just the first step from managing multiple spreadsheets and floor plan diagrams to getting a consolidated view with a single version of the truth is a small step with huge benefits.
Layers of the model build on each other as you progress:
Phase 0 – Managed chaos: This is where most companies are today. Dealing with multiple spreadsheets, one each for servers, power cabling, network cabling, rack space and power consumption. They are often out of sync with each other and with the reality on the floor.
Phase 1 – Information and application consolidation: Consolidating your spreadsheets and diagrams into a single application. Entry-level DCIM products can import spreadsheets directly into their database. From there, users get visualizations of the data center floor and individual racks. At this level, users should also be able to see available space and power, get snapshot reports on precise utilization at a single point in time, simulate rudimentary “what-if” scenarios for deploying individual assets (like servers) and perform basic move, add and change (MAC) management.
The organization and consolidation of the raw data in Phase 1 significantly increases accuracy of the data center model. As a result, managers can make better decisions and reduce human errors. Making the jump to Phase 1 is simple and takes from a few hours to a few days.
It is crucial to get a strong Phase 1 deployment, however. If you have “dirty” data in your DCIM database, everything you layer on top will be based on faulty assumptions. It is important to build on a solid foundation.
Phase 2 – Process optimization and historical reporting: The next phase of DCIM maturity requires a greater commitment to changing your management processes. This is where data center workflow processes are programmed into the DCIM software and where the software is used to reinforce them. This level is also where you can begin tracking data center utilization rates over time. It allows for better forecasting based on historical results. Some managers chose to inject real-time information into their DCIM software at this stage, increasing precision of the database.
Phase 3 – Strategic data center planning: This is where you begin performing multiple “what-if” planning scenarios with large numbers of servers, using the DCIM solution to identify potential failure points in the power chain and using predictive analytics to fully optimize the use of power, cooling and space. This is also where auto-discovery software can be used to automatically audit devices in the data center and identify potential errors in the DCIM database without a manual audit. While integration with other IT solutions and processes often happens in Phase 2, the third phase is when the DCIM solution should be integrated with helpdesk, configuration management database (CMDB) and other IT management software solutions.
Final word of advice
One very important factor to consider when choosing DCIM software is to look for a product line that allows you to move seamlessly from Phase 0 up without changing the software platform. Staying on the same foundation makes it easier to progress up the Maturity Model.
There are many DCIM solutions in the market that only offer products that don’t go beyond Phase 1. This means you will be forced to change vendors once you are ready to progress. Similarly, there are vendors that only offer enterprise-class solutions that make you deploy Phase 2 and Phase 1 aspects of the Maturity Model simultaneously. The problem with this approach is it forces you to make too many changes to your data center processes too quickly before letting the foundational aspects of Phase 1 get settled. This puts your DCIM deployment at a higher risk of failure.
The key to success is to start small but think big as you move up the model. You can move through the steps quickly but never in parallel. Like most IT projects, you want to take DCIM step by step, adding layers of sophistication gradually. This will increase the likelihood of a successful implementation and demonstration of the value a well-run data center can provide to the entire IT organization and enterprise.
About the author: Todd Goldman is VP of product management at nlyte Software