Lately, I have participated in a number of online group discussions which focus on trying to define what DCIM is. There is clearly more than one definition circulating from various sources. Each of the major analysts have their own (similar but not identical) definition for DCIM as well. What this does is to create a level of confusion that end-users are forced to wrestle with in their search for answers.
To create a generally agreed definition about what DCIM is, I think it is worth starting a discussion about WHO could use the features found in the industry’s current DCIM offerings, along with a special notation on where DCIM will be going over time (likely to include orchestration & automation). I think in general many DCIM vendors have done a poor job articulating the problems they are solving, rather than just the features that they deliver. Simple mistake right? Well, no. DCIM is not a one-size-fits-all type of technology. My contention is that to put together the list of problems that we are solving, we need to step back a bit (just a small step) and look at the roles of the people that we are solving these problems for…. THEN we can look at the problems each of these people have, and ultimately each vendor can then map their features to one of more of those problems. The end-user community will welcome this clarity!
So let’s start with the TEN faces of DCIM. Here is my list of the TEN roles/personas that I see benefitting from DCIM, and my view about why they would benefit from DCIM:
- CFO. This may be the most important user profile to account for when considering the purchase of DCIM offerings. In general, the CFO wants to know that his/her investments are well spent, that the investments are cost-effective, and that the investments are all focused on providing just the right amount of service, at a low risk. The CFO wants to be able to get a pulse of the operations of IT, and look at fiscal KPIs such as “cost per transaction”. These financial folks see the DCIM solutions in direct support of the accounting side of IT, and want to be assured that the cost side of IT is driving behaviour. The technical challenges can easily be managed with unlimited budgets, but those days are long gone.
- CIO and/or the VP of IT. This is a related role to the CFO, but the CIO knows that there is an almost infinite number of combinations of technology that can be deployed to meet the business needs, but that these combinations change over time. The modern CIO needs to be assured that DCIM will provide them a view of lifecycles for deployed solutions, and that each and every piece of gear that has been installed is needed, supported, managed, and planned… all at the minimum cost to meet his/her committed levels of service.
- Executive Team. Consider the executive team as the At-A-Glance members of the DCIM user population. These executives tend to want DCIM to net out the results or issues in summary fashion. Roll-ups of costs, costs per unit, overall status of production, etc. DCIM provide one of the simplest ways for the executive team to understand in near-realtime the performance oftheir $Billion dollar investments. The executive teams also looks for DCIM to provide grande-scale trends, such as the services availability, processing or total energy costs year over year.
- Data Center IT Manager. Much closer to the hands-on side of DCIM, the Data Center IT manager is chartered to operate the deployed solutions in an ongoing fashion. Their primarly concern is consistency in performance, efficiency of processes, and the repeatability of best practices. The Data Center IT Manager is tasked to assure that the services that have been designed are up and running at the required level of service. DCIM becomes process enabler for this role, and workflow associarted with change is of primary concern.
- Facilities Manager. One of the initial supporters of DCIM for most corporations, the Facilities Manager considering DCIM is looking to augment their existing BMS/BAS systems with modern updates. He/She is looking to add more granularity and analytics to power and cooling systems that have traditionally been fairly macro and static in nature. DCIM presents an opportunity for Facilities leaders to build on top of their rigid power and cooling structures, with a goal to reduce costs and reduce risks. Facilities folks are looking for more detail, more control, and more capacity planning.
- IT Business Analyst. Here is where DCIM is utilized as a vast repository of information combined with various means to filter the data in hopes of establishing business trends and optimizations. The business analysts may consider a data center as a compilation of technologies, servers, storage and networks or they may wish to view it by by something else altogether, such as business unit, warranty status or project types. The business analyst is looking for DCIM to provide a data warehouse with all pertinent asset and resource data, and then the ability to easily naviate and filter that knowledge to create actionable derivatives. The analyst looks at costs, and lifecycles, and value as their mainstay.
- Structured Cabling team. Another hands-on team that looks to DCIM to document and manage the change that is associated with connectivity. In a modern data center thousands of connections are required for data to connect to other systems, storage and peripherals, and each of these needs to be tracked and actively managed. They look to DCIM to clearly identify each of these connections, and provide that understanding when they are being asked to make changes. Cabling can quickly become fairly unwieldy, so DCIM can be one of their best bets to document with a high degree of accuracy what has been built, and to support work-orders for changes to this connectivity to be made.
- Data Center Technicians (deployment and remediation). Perhaps one of the most eager users of DCIM, these technicians are constantly making physical changes to the data center, and DCIM provides the formal structured support mechanism to assure those changes are well understood, executed quickly and accurately as needed, and fully documented to assure compliance and ongoing operations. The technicians are faced daily with installing, replacing or removing gear or making connectivity changes, so work-orders are a main staple of the job. DCIM offers the ability to provide clear guidance and step by step instructions on what to do and when. In a typical data center, these technicians execute thousands of work-orders per year, each one a mini-project, and all of them overlapping in time, so the tools to do this quickly and efficiently are paramount for a corporation’s success.
- Asset Manager – These users have very specific needs associated with the fiscal management of capital assets. DCIM becomes one of the best means to assure that a clear understanding of the lifecycle of those assets, where each is installed and its value to the organization over longer periods of time. The asset manager is chartered to assure that devices are in-service while their value to the corporation is high, that each device is supportable and that maintenance plans (including warranty status) exist. Previously a spreadsheet user, DCIM offers a realtime view to the asset manager to allow an accurate representation of the capital and operational expense costs associated with data center gear at any point in time.
- Capacity Planners and architects – One of the most visible users of DCIM, their role is to determine the operational needs for the data center now and into the future. Capacity planners and architects work together to assure that the overall systems works, the right amount of resources are available, and that fixed limits are well understood well before they become a problem. Power or space for instance are fixed resources with huge investments required when current levels are exhausted. The time frames associated with adding these new resources can be more than a year, or in some cases require entirely new facilities to be built elsewhere. These planners use DCIM to project when and how resources will be consumed over time and identify potential problems well before they impact the business.
It is clear that each of the different vendors’ DCIM offerings address a different set of these users’ needs. A monitoring package might be interesting for the facilities manager, but it would NOT help the Structured Cabling team at all. The ability to visually navigate through the racks and rows would be of great help to the Data Center manager, but wouldn’t help the CFO and so forth…
So the key is and will always be to identify just WHO your constituency will be, short and long term. Who are the most beneficial users that should be involved in any DCIM initiative? Only when that full set of people is identified, will YOUR definition of DCIM become clear.
Trying to plan a data center that can accommodate needs for 10 years has always been a challenge, with the classic fallback being to just overbuild everything to allow for any conceivable future.
But when the economy tightened, organizations began to re-think this practice and have become extremely conservative in bringing new capital-intensive future capacity online. Each in-progress project is being scrutinized for scope and priority, and existing data center structures are being studied to determine how they can be better utilized immediately.
At the same time, all professional IT organizations are spending a lot of effort to create a set of best practices and implement solutions which assure they understand their business computing needs and can articulate their computing capacity demands over longer periods of time. The goal is to closely align the supply of data center capacity to those needs.
One of the industry’s newest technologies to address this business planning function for data center capacity planning is the Data Center Infrastructure Management (DCIM) Suite. DCIM Suites provide a set of capabilities that enable the monitoring and modeling of everything in the data center for the purpose of capacity planning and operational support. Every device, every connection, where they are placed and how much power they draw.
Read the whole article over on CIO.COM
VMware’s annual extravaganza, VMworld was held a couple of weeks ago in San Francisco and has grown to more than 20,000 attendees. This is a really big deal for the industry and VMware has the right to be proud of what they have done. Ten years ago they seeded the industry with commercially viable dynamic capacity and these approaches have now been adopted to varying degrees by all Enterprises worldwide.
The VMworld attendees ranged from CIOs and CFOs, to System Administrators and Programmers. This wide diversity in job function is no surprise as the range of topics being discussed in great detail also ranged from the Business and Financial planning for IT, to Disaster Recovery and Planning, optimization strategies for dynamic capacity and everything in between. The conference also sported a huge trade-show floor, chock full of vendors who had really tuned their products and messages to embrace the world of virtualization. Not just virtualization of servers, but virtualization of storage and networks, and even the hybridization of computing to include the Cloud. The whole event had a core theme about modular capacity and how to add capacity in realistic building-block increments. 250 vendors packed the trade show floor and spoke directly to the audience’s capacity needs. It was impressive.
Amongst all of the exciting vendor demonstrations and discussions, one vendor in particular stood out by carrying a message not previously seen before in a conference like this. It was (wait for it)… Nlyte Software. Nlyte was telling a story about the coordination and placement of virtualized load upon physical resources. Nlyte was the ONLY vendor on the show floor telling this often overlooked story, and by the looks of people visiting their booth, it was very well received. As a pioneer in the DCIM category, they are masters at managing the physical layout and connectivity of everything in the data center. They manage the lifecycles of devices from the time they come into service all the way through their decommissioning. Nlyte’s story as it applies to the virtualized world is straight-forward and not contrived- as workloads are being moved around dynamically, it is critical to coordinate that movement with the underlying availability of resources needed to power and cool those devices. The technical discussion about why this is so critical centers on the power consumption curves of devices as work is applied to those devices. In simple terms, the more work being performed, the more power consumed and the more heat which is generated. Allow this automated movement of workloads enough to the same part of a data center and it is very reasonable to expect catastrophic failures in processing due to inadequate power or limited cooling capacity. This is where Nlyte comes in. Nlyte provides the coordination of dynamic placement of virtualized Guests with the underlying resources needed to power and cool their Hosts, and does this in real-time.
Virtualization is clearly not only here to stay, but has now be applied to storage and networks. All of this abstraction is focused on decoupling the hardware from the software which is a great thing for application developers and service delivery, but creates an critical requirement to actively manage the physical layer as a resource of the processing itself, much like CPU cycles, memory or disk space. Gone are the “Good ‘ol days” where data centers are being over built, over-provisioned and over-cooled. In those days, everything physical seemed ‘infinite’ in nature so not action was needed.
Today, those resources have actual costs which are non-trivial, so data center ‘right-sizing’ is the standard practice. As data centers are right-sized, the resources must be coordinated with the workloads. Abstraction is a great thing, as long as the fundamental foundations are not forgotten.
Anyone that has spent any time with an Nlyte representative would likely have heard us say that one of the primary values we bring to the industry is to support the massive change management requirements in the data center and that Nlyte supports the business optimization associated with that massive investment over the long term. I know you would have heard those words, because that’s what we do. Sure we also handle monitoring of power and drawing high-fidelity elevations of racks and floorplans, but the BIG value is managing asset lifecycles over long periods of time. It’s possible however that you (and perhaps your peers) would NOT have had the context to internalized that message in your world, as it applies to YOU! Let me explain a few Data Center numbers to help us come to a common communication ground.
First, a Data Center is expensive when viewed in the context any point of reference. A conservative estimate for a ‘Tier-3′ structure and all of the facilities inside it (power, cooling, lighting and security) is about $1000 per square foot. So a 50,000 square foot center will cost about $50MILLION to build. Then you have the challenge of filling it up with gear. Every center’s computing needs are different, and gear can range from simple 1U servers to advanced 15U blade centers. And all of those ‘U’s need to be mounted in racks. Lots of racks which take up that finite space in the data center. So how many square feet is a rack? 32! Yup. Every rack in your data center consumes about 32 square feet for estimating purposes. (The rack itself, and all of the maintenance clearances, the walkways and facility/overhead space. Trust me, if you think long and hard about all of the ‘overhead’ space used in a data center your representative number will be closer to my estimate of 32 than it is to a low ‘gut’ figure of say, 10 or 15). To complete the picture, a good estimate of the number of IT devices per rack is 25. Sure, some racks are more dense, some are less, so we use a figure of 25. Now using a little hand-waving and estimating the price of various types of gear, I think it is very safe to say the cost to deploy gear inside the data center will be about 3-4 times what it cost to build the structure itself, or about $3000-4000 per square foot, which yields about $100,000 per rack for the equipment itself. (I have seen single filled racks costing $200,000 or more as well, so the figure of $100K is just a mid-point estimate). How many racks will you have? Simple math again, 50,000 divided by 2 (half filled), and then divided by 32 (per rack) yields about 800-1000 racks.
So now you’ve built that 50,000 square foot data center, which cost $50Million for the box, and you’ll spend another $100 Million to fill just HALF of the space with IT gear on the first day. (HALF is a conservative estimate of your day-1 usage of the space). Lastly, remember that each rack has about 25 device on average, so your shiny new data center has about 20,000 devices installed on the first day it is operational!
But HERE is where the story gets more interesting strategically and much less intuitive…
As it turns out, for the past 20 years your finance team has been biting their lips and clenching their teeth when they look at the data center general ledger. First of all, it has traditionally been a sunk cost, which means that every dollar spent for the data center works against the bottom line. These same finance folks have also lived with the burden of aging assets on the books way past their financial schedules, and dealing with assets that are ineffective and yet still operational. In a nutshell, in a highly optimized data center planning model, most gear is designed to ‘live’ on the books and on the data center floor for 3 years. Yes, 3 years. Most depreciation schedules, leasing schedules and even warranty schedules are built upon this three year model. So what happens after three years? The gear starts COSTING the company much more than it needs to and it’s actual VALUE to the company sharply decreases as all of the financial benefits vanish. (Not to mention that the technology itself is also likely at least 75% slower and this old slow gear is using 50% more power than something new). While in the old days it was ‘cool’ and a bit of a right of passage to prove that equipment could be keep alive for years and years, it is no longer the business model that is rewarded. Gear needs to move into service quickly, be consumed fully when it’s value is high, and then be retired aggressively when it’s schedule dictates. In short, IT gear that stays in service past it’s three year schedules directly work AGAINST the bottom line efficiency of that data center.
So what does this mean? For your data center to be optimized and run as a strategic asset, ONE-THIRD of everything you see in the data center should be changed every year. In that example above, one third of 20,000 devices or almost 7,000 devices per year. Now with 200 business days in a year that means 35 devices each and every day SHOULD be changed. 35 new ‘projects’ each and every day, and anything that doesn’t get done today becomes backlog for tomorrow. This is not just theory, but the new reality of running a data center as a strategic asset.
In the new economics of the data center, these simple and yet strategic math discussions are becoming commomnplace with the executive teams, and beginning to perculate down. This type of financial management of the data center as an asset is real and here today, and this is what Nlyte does. We manage hundreds of concurrent projects and workflows to enable those change projects to be on schedule and accurate. We support the moves, adds and changes and we identify each various device’s business schedules. DCIM is a fascinating business management solution, and the Nlyte implementation in direct support of IT’s financial goals is a hard value to overlook.
We’ve been busy at Nlyte meeting with end-users that are now searching for mature DCIM solutions that can be quickly deployed. Together with Server Technology and RFcode, we conducted half-day meetings in 7 cities in the past 2 months. Over a thousand people registered to join us for these meetings.
The meetings allowed the three vendors to share Data Center transformation and DCIM ideas, present current capabilities, but most of all, we LISTENED to what was happening across this statistically valid cross-section of the industry. With that, I wanted to share some of our observations on topics that had enough convergence to identify a trend or direction.
The top dozen topics we heard:
1. Change is much more common than it might otherwise appear
The vast amount of change, which continuously occurs in all data centers, is deceptive. While it appears to many that the data center is fairly static, in fact due to refresh cycles, warranty schedules, remediation and new applications, a quarter or more of a data center is actually changing every year. Each of these changes is a ‘mini’ project and must be managed and documented in the context of all of the other projects being planned. Many of the attendees had no idea that so much change was happening ‘behind the scenes’. Management of all of these projects is daunting and attendees agreed that it simply doesn’t happen very well, which one of the reasons they are searching for a DCIM solution!
2. IT and finance organizations are natural partners
Nearly all of the attendees were from the IT and Finance organizations. It was clear that their counterparts in the Facilities organizations already have a wealth of tools that maintain HVAC, Lighting and Security function and it was the general consensus by attendees that these facilities systems already run fairly well and make would a perfect source of energy-related knowledge towards the goal of making better business decisions for the whole data center. Many attendees discussed the need to actively manage the data center as a complex business asset with many components, each requiring pro-active planning and operational efficiency. A related topic that arose was the need to enable cooling resources to become more dynamic and responsive to granular demand, but that topic was not investigated deeply during these discussions.
3. Integration and customization important but more ‘off-the-shelf’ is best
Integration and customization was a hot topic that has real strategic and financial impacts. It was generally agreed by attendees that the percentage of a solution that is off-the-shelf directly affects their willingness to adopt any newer technology. Most attendees acknowledged that some professional services were expected, but they felt like any more than one-third of the project would be prohibitive and cause their deployment to be locked into specific versions. It was unanimously considered a good thing vendor DCIM vendors get together to create ‘known working’ solutions that integrate deeply and leverage combined value well.
4. Temperatures on the rise in the data center
Increasing the standard temperature within the Data Center was a common goal, but most data center operators are ill prepared to execute that strategy due to lack of granular visibility and on-going concern for outages. The discussion about raising data center temperature quickly turned into the need for limits of those potential changes, and how those changes could affect power consumption. Most attendees are still running their data centers at about 68-degrees. It was agreed that higher temperatures are a good goal, and most could see running at 72-degrees without too many changes. Each attendee wants to drive their data center temperature up within the next year, and even higher temperatures would be considered only after some experience has been attained.
5. Intelligent power and environment monitoring moves up the chain
More than half of the audience has data centers with simple ‘dumb’ power distribution devices and very few (if any) network-attached environment sensors. Unanimously, each attendee will make the choice to move to or grow with intelligent power chain and environment monitoring components when the opportunity presents itself (new, reconfig, etc). Most attendees can see the need for these energy monitoring solutions as part of their new management plans. A lively discussion about intelligent asset locator technologies was well received and attendees agreed their current manual audits to determine asset location were less than desirable. Most are looking for a strategic approach to solving the asset location problem.
6. DCIM is a priority, but the definition of DCIM is far too broad today
Funded DCIM projects exist in nearly all attendees’ plans for 2013, however the definition of what ‘DCIM’ entails varies widely from attendee to attendee. Ironic because funding and a genuine desire to take SOME FORM OF ACTION exists with most attendees today. Most are currently in the evaluation of DCIM solutions and have a plan to deploy intelligent power, monitoring and a DCIM suite within the next year or so as new data center space comes on-line. Attendees noted that dozens of DCIM vendors approach them over the course of the year, and each has begun to sound the same. Attendees are looking for non-biased DCIM guidance.
7. No more business as usual, the modern data center is here
Attendees are looking for Data Center leaders who understand the ‘new’ challenges in the data center. They suggest that lots of folks can be found to discuss the data center of 2005, what they need is state of the art experts. They are searching for highly skilled talent in their new hires and are looking for new vendors with new offerings to help craft their next data center. Business as usual will not be the strategy going forward and many of the attendees have already had to re-tool their own knowledge to participate in their company’s IT structure plans going forward. Their future will likely be a changing combination of In-House, Co-Location, Modular and Cloud services.
8. Energy usage and capacity planning are the highest priorities
Understanding energy usage and creating the business processes to actively do regular capacity planning were the two highest priority needs identified by the attendees, followed closely by compliance and audit needs. While most attendees acknowledged that the rising cost of power has become the catalyst for change (which raised the visibility of the whole data center operation), it is their belief that energy savings is just the tip of the iceberg and will easily be dwarfed by saving available through the adoption of streamlined and highly efficient operations. Ultimately, creating a new data center plan which is efficient, defendable and supportable was the attendees’ major goal.
9. Knowledge is key across IT, facilities and finance departments
Most attendees were not looking for a single tool to create a literal ‘single-pane of glass’, but instead were looking for a way to access the most complete store of knowledge to make better business decisions. It is highly desirable for the multiple available sources of knowledge from the IT, Facilities and Financial systems to converge in their business support processes. Today most attendees agreed that the financial aspects of the data center have been ‘buried’ for far too long, and they are encouraged that the data center was beginning to be treated as an entity that must be planned and managed, with general-ledger style accountability.
10. Energy usage reports by function and organization are in high demand
About two-thirds of the attendees did not possess the knowledge nor any formal means to apprise their executive teams with current or future energy demands and the costs that would be associated with that demand. In fact the executive team has not even requested this level of detail in many cases. That said, the ability to report on asset energy usage by location, by business unit, by function and a wealth of other views would be highly desirable from a DCIM solution.
11. Education is needed around cloud computing for DC planning
Most attendees have heard about Cloud and Modular styles of computing, but most were unsure how each pertains to their existing in-house and co-location planning. There is a general awareness that there will become a hybrid model of computing across all four styles, but knowing which business applications to put where will become a detailed balancing act and one that the attendees feel that are ill prepared to execute today.
12. IT and finance need to come together for cost/ROI planning
Nearly all attendees said that one of their top initiatives for 2013 is defining and measuring their on-going IT projects in terms directly relative to COST. Not merely ROI, but actual costs associated with doing a unit of work and the long-term impact of those investments. In addition, most IT professionals in attendance were still unaware of the true financial impacts associated with the accumulation and continued usage of older IT equipment well past its typical 3-year warranty and depreciation schedule. Many felt that IT and Finance organizations need to work more closely to provide a baseline.
We will continue to be teaming with DCIM components providers like ServerTech, RF-Code, FieldView and others for additional face-to-face meetings to demonstrate the DCIM eco-system which really is forming for the strongest of players. Stay tuned for more details.