Top 5 Data Center Tips for 2017

Featured Image

Data Center Industry Roadmap 2017-2020

Data center design, build, operation and ownership are entering a new era. It will not be long before those operators who failed to adapt, will perish by the virtue of their own making. Whether there will be more or less data centers, greater or fewer outsourcing, expansion of public or private clouds, larger or smaller data centers, are each subjects of another discussion and another blog. But to build a data center roadmap that is necessary to deal with the recent challenges and still remain a prospering business by 2020, all data center owners must align themselves to the below in 2017.

1. Become application centric – STOP leaking…

If you have been a fan of Cisco, HP, APC, Emerson, etc., those days are over. All brands and goods out there should become commodities that you would easily be willing to alternate and exchange, if doing so makes business/financial sense to you. Data centers must stop building around a preferred technology. Rather, understanding the application and building everything around it is key. This should be your #1 priority for 2017. You must find relevance between all the seven stack layers of your Application Ecosystem (AE) and design to suit, build to suit, outsource to suit, operate to suit to one thing: “the application”.

Application Ecosystem (AE) 7-Layer Stack

How effective, safe, secure, resilient, robust, and efficient you build and operate your application ecosystem is vital for you to successfully face and address the new breed of data center and information technology challenges. It is only when you think application is king that you would realize how much you are bleeding and leaking either across or between different layers of your application ecosystem. For some organizations the misalignment between different layers and subsequent leakage of resources could mean billions, while for many it would be in millions. But whatever the amount would be, it will be too much to sustain if you don’t rollout this initiative in 2017.

2. Educate your people – GET serious…

If you have been limping on one leg so far with regards to your human resources, get serious. You can’t possibly remain competitive or financially feasible over the coming years, should you continue this path. Your people are the sole reason you either win or loose in the data center arena. Whether you run your own data centers or you outsource them completely, the individuals dealing with your “data” must be trained, period. Their value judgment can give you the right information as decisive inputs, manage and operate your infrastructure effectively, avoid mistakes and prevent human errors, choose the right products/services for your application ecosystem, understand and communicate effectively with their peers in the same layer or layers above and below their own AE layer, and ultimately prevent massive wastes, which are probably ongoing as we speak. Optimizing the output of your data centers goes back to how well-informed your people are about their role and its cause and effects on your application ecosystem. You must give them that education and mandate them to learn. If your people think data center is simply the housing of IT systems, warehouse of chillers, generators, structured cabling, core routers and switches, you can already see the tip of the iceberg of problems coming your way. Do NOT be too confident with your staff’s capability to run specific components. How to run a generator or configuring a router, should be the prerequisite for one’s job enrollment to begin with. How to run all of that within the context of data center is another discipline all by itself. It is the interoperability of all these components and more that you need to educate your human resources on. Earning such discipline would not a natural occurrence, at least not on your watch. Data center discipline must be learned, hopefully not be trial and “error”. If your staff don’t step up and understand the holistic nature of things, sky is the limit for how many facets you would be loosing from, practically on daily basis. You cannot afford to walk into the upcoming generation of data centers without the appropriately educated personnel.

3. Become industry specific – BE yourself…

If you are a healthcare institution, your data center must look different from an airport’s data center. If you are a university, you should have a unique set of differences with the way in which you design, build and operate your data center as compared to a defense or military organization. This analogy is applicable to all industries. If you are not adhering to your specific organization not only from a compliance standpoint, but also from the need and approach perspective, you are missing the point. You must identify the uniqueness of your specific industry, which could range from the people you would hire, the outsourcing modes you would consider, the geographical regions you could have presence, the type of telecom you can utilize, the average density for your cabinets, the safety and security precautions you must recognize and much more. Thus, your data center should have a very definitive look to it that is specifically unique to and aligned with the nature of your industry.

4. Become Cloud driven – ZOOM out…

Instead of building a 2N+1 facility then thinking how it would serve within the whole scheme of things, it is only wise to devise the resilience and capacities you require within your cloud. You could have a number of N-based DCNs that together deliver you a much more robust infrastructure that are easier to commission, decommission, manage, etc. than having one gigantic and fault-tolerant site that hosts all your eggs in one basket. After all, that “one site” is a single-point-of-failure. You could have two 2N sites, or one 2N + one N+1 site that give you all you need and for your specific requirements. You might find that cheaper to run and maintain than having eight locations to deal with. The pros, cons and feasibilities of all of this will manifest into the cloud mindset, not the narrow site-minded view. You can only warrant true availability, security and efficiency when you zoom out of a specific site and see your infrastructure from a broader spectrum. In short, you should care less how many generators are running in a given site if your overall cloud architecture is satisfied for a particular site to just have one. If you stepped into 2017 without a grander cloud initiative, 2018 will be a tough one.

5. Build effective manuals – DOCUMENT everything…

You buy a bicycle, a smart phone, or anything for that matter, it usually comes with a manual. Devices run better, and lot of mistakes could be avoided, when their operators read manuals first. This was a lesson learnt at the break of the 20th century and implemented throughout World Wars I and II. How can we run mega facilities that cater to lives in the 21st century, without proper manuals and effective documentation? Without documentation every new rollout, every installation, every new hire, every vendor, task or management call is a challenge. What you need to do is to make sure every task becomes a routine and falls within a refined framework with proceeding guidelines that have already been thought of and planned for ahead. If you are running 10 sites today, you need minimum 11 manuals. You would need, at the very least, 10 manuals for each specific data center node to address the needs and concerns of the resident as well as the remote staff that deal with a specific DCN, making them aware of exit routes, UPS maintenance regime, and anything that is node-specific. Then, you need a universal data center cloud (DCC) or topology manual that tells your cloud operators how the whole setup is interlinked, managed and run and how is application being deployed and delivered across these multiple DCNs. Matters of latency, weather forecasts, global installations and operations, training requirements, SLAs, OLAs, application and compute management from the aggregation of all data center nodes must be addressed in the DCC manual. Remember it’s all about efficiency. Therefore, make sure your manual is neither too thick nor too slim. In either extreme, it will loose its effectiveness. It must be sized, worded and toned just right to aid deliver the KPIs it aims to preserve.

At the brink of 2020s drastic changes can be predicted. One can only be prepared for future if the challenges of the past have already been resolved.

At the brink of 2020s drastic changes can be predicted. One can only be prepared for future if the challenges of the past have already been resolved. The above five action items are the most basic and fundamental adoptions data center stakeholders must make. For those who have a vision of what’s ahead, 2017 should serve as platform for a set of phased initiatives. Such initiatives must cultivate and render a 3-year (2017-2019) agenda for change.

List of Comments

No comments yet.

Leave A Comment