For whatever reason, it is time to get out of your data center facilities and do either a data center decommission or data center migration. You as the data center manager, have been tasked with getting all those big hunks of metal out of the computational behemoth that is your data center.
The good news for you is that you have options. Not like newly single Justin Timberlake level of options, but you’ve got some choices for how you want to play this.
Your facilities can be:
1. Replaced by cloud services
- Fast and cost effective
- It’s super appealing to data center management to be able to plug and play apps/switch over seamlessly
2. Replaced by colocation services
- Security, support, flexibility, reliability, cost savings, disaster recovery.
3. Consolidated into other data centers
- Management may have decided they want everything in one spot. Maybe a Chief Technology Officer decided he doesn’t like flying anymore. Whatever the reason, you’re moving stuff from one into another and getting rid of redundant parts.
4. Decommissioned entirely
- Because nobody wants them anymore in any capacity. Forever alone.
Regardless of the data center’s destiny, you can find savings and streamline the process.
Cloud, Colocation, Consolidation or Decommissioning?
The process will differ depending on the real estate situation. If you are leasing the building, you’ll need to review the lease contract for any requirements on decommissioning and the final state of the building. If the building is owned, there are a few options. If the data center is within a building used for other business functions as well, it can be converted into additional office space. The space could also be leased to other companies if cooling and power specs meet standard thresholds. If they don’t meet standards for modern cabinet densities, then generators and other infrastructure equipment will likely have value for other businesses.
More often than not, data center decommissions are part of a larger project to relocate equipment to a different location. When done correctly data center migrations can be quite reasonable and have minimal costs for data center management. When implemented poorly, the results are far from ideal for the beloved bottom line. The following section reviews how to prevent that.
Data Center Migration
At this point, Data Centers are popping up faster than weeds in the garden. Couple that with the constant consolidations of old data centers into main central locations and you’ve got a nice recipe for very frequent migrations in our wonderful world of IT.
And for data center management, the word “migration” is enough to give many nightmares with good reason. Transitioning apps and data over to a new location can be quite the grueling endeavor…which is about as much of an understatement as saying that Tom Brady is decent at football. It will never be easy, but there are ways to make sure that your data center migration isn’t more of a headache than it needs to be. There is an important distinction to make with migrations: those to the cloud, and those not. Let’s go through them.
Cloud Data Center Migration
It seems like the cloud is blowing up faster than fidget spinners! Some major cloud providers are expecting to triple their infrastructure by 2020. Why? The reason is simple: unlike fidget spinners, the cloud is awesome and offers a lot of benefits.
There are some limitations with cabinet hosting that make transitioning to the cloud so appealing. Usually, physical environments are over engineered just in case the workload demand goes crazier than a soccer mom shopping on Black Friday. This complicates data center management immensely. Now you’ve got a big chunk of servers just sitting there, completely unnecessary most of the time, pretending to be important. Kind of like that guy from the support desk who’s actually just browsing memes on reddit. Not very helpful.
Can’t Predict the Future Needs of Your Data Center? No Problem!
Predicting exact workload requirements is like trying to hit a moving target in warp speed. The cloud’s great because you don’t have to predict workloads in the same way. You just pay more for more computing power until things slow down again, as opposed to buying more servers, memory, and drives that you won’t even need in a week. Implementing cloud services even incrementally can greatly simplify the data center management process.
While migrating to the cloud as opposed to another location has less moving parts, here are some tips to make sure it goes smooth as ice.
1. Choose the platform for your needs
Choosing your platform (i.e. OpenShift, VMware) is going to be the first course of action.That decision is usually based on cost and functionality. Then you realize that providers can be as different from each other as vegetarian falafel is to actual meat. The big takeaway here is that you should shop around. A very complex platform without rapid support wouldn’t make sense for an enterprise that has limited experience and is prone to frequent breakdowns.
2. Don’t skimp on discovery
It’s vital for data center management to be extensively aware of these steps before a cloud migration: License portability, app compatibility, any custom development work, compliance and security concerns. Considering each step is important, as The last thing you want is to realize your most mission critical app doesn’t even function in the new platform at the end of your data center migration.
3. Test these before you go live
“Test before you go live” seems like easy advice, though it’s easy to underestimate just how much attention to detail this will take once you get underway. Test your latency, test how stable your apps are, test for bugs. Can you be certain that your new resources can handle expected maximum workload demands?
4. Phase, don’t dump
If you move over the less critical apps first, you have more time to look for any issues while the stakes are low.
5. Communicate with end users
Standard procedure when your favorite website goes down is to freak out and mash the refresh button. If you tell your users that downtime will happen ahead of time, they’ll probably still mash the refresh button anyways, but you’ll mitigate some of the backlash.
A migration to another physical location, sometimes also called a relocation, is logistically a very nuanced endeavor. We’ll go over best practices for these migrations next.
Non-Cloud Data Center Migration
Data center relocations are fraught with potential logistics problems; any time equipment has to travel, the risk of damage en route and loss will always be present. To avoid those bumps in the road, the following are effective mitigation strategies.
1. Acquire a specially assigned project manager:
It’s quite likely that you have a perfectly competent and capable data center management team on hand. However, a data center migration presents a massive undertaking. Would you assign a general practitioner doctor to an extremely nuanced heart operation on the president of your company? More than likely you’d want a heart surgery specialist to be on the case. A data center migration is just as complicated. In fact, it’s probably more complicated, but nobody’s life is on the line.
2. Determine what you’ve got
Before an enterprise considers a data center relocation, it’s vital for data center management to do an extensive site audit and take stock of inventory. With a complete picture, you can determine which systems are machine critical and should be moved first, and which systems have mutual dependencies.
3. Document to dominate the relocation
With relocations, the poignant emphasis on documentation can throw plenty of techs off guard. Many are accustomed to keeping incredibly important information “in their head.” If you’ve ever heard the phrase “fail to plan and you’re planning to fail” if you fail to document, you’ll be documenting a million dollar mistake soon enough.
4. You need to have four documents:
- A detailed account of what you have in its current state. Storage needs, hardware, software, dependencies, networking details, equipment lists, support processes.
- A detailed account of exactly what an ideal outcome for the relocation looks like. What conditions are you going to use to define your success? This account should include any expected updates in the future like storage upgrades or virtualization.
- All of the steps necessary to get from point A to point B. Budget out each step, and its necessary components individually. Elaborate on each process needed to finish the relocation separately. Detail any known risks and their countermeasures, all expected hardware/software needs, a plan for communications, a detailed timeline, and the effects of end user processes on your design.
- An hour by hour plan with every date, step, responsibility and person responsible, interaction, linkage, and general move event.
4. Don’t lose at logistics
Once all of those documents are prepared, success is almost entirely determined by the logistics during implementation. If qualified people are in the correct positions, supplied with adequate tools, at the appropriate time, success is imminent. To that end, getting solid teams in place, such for decommission , packing and transportation, unpacking, and installation, stocked with the best people by a logistics expert is the natural next step for data centerment management to take.
5. Don’t underestimate necessary resources.
Decision makers typically grossly underestimate the complexity of a data center migration. The operations teams executing the relocation with limited resources could tell ahead of time that success is not guaranteed.
To that end, it’s better to have a nice mix of those who know their way around a data center migration, and those who know your data center environment. Engaging a vendor with expertise in relocations to bolster resources is a common practice that typically pays dividends for smoothing over the transition.
6. After that’s taken care of, establish what tools you’ll be using to oversee the process.
Ask yourself if you can get away with the basics like Excel, Microsoft Project, google docs, etc. Generally that’s like trying to fight an elephant with a BB gun, but if it works more power to you. Data center migrations are basically organized disasters. Make data center management as easy as possible: think about getting systems in place built specifically for your purposes.
For example, every step of your migration can be orchestrated ahead of time with runbooks that detail every second of your data center migration. 2017 is a magical year for data center management.
7. Be like the government response team in 24, get yourself a command room.
Email and conferences will inevitably waste time as all the back-and-forth adds up. With one centralized location to enact any higher level operations, visual dependency maps can be a huge lifesaver here. Visual maps can give cues for further investigation, such as a lone system not connected to any other systems. This is suspicious and would potentially have been a missed error if data center management used traditional methods. A real time infographic is often preferred over scouring thousands of spreadsheet lines, although there are a few outliers that prefer the latter.
8. Equipment should be categorized based on whether it needs to be updated, replaced, or decommissioned.
As much as everyone seems to hate labels in 2017, there is a time and a place for everything and it will make the move go much smoother.
9. Consider hiring a data center migration expert to audit your data center and give you some feedback.
A little constructive criticism never hurt, and this is important to get right.
10. Review service agreements and contracts with current vendors.
Nothing is more fun than moving only to realize your contract is voided and they won’t service anything. A data center migration is the perfect time to ”trim the fat” so to speak, so this could be a great time to fire any partners or vendors that you’ve been unsatisfied with.
11. Double check that your new location can accommodate all of your equipment.
Look for any compatibility problems that may arise. Each data center location will have different restrictions, so evaluate the differences and their relevance to you.
12. Plan your data center migration on a day that is expected to see low traffic and won’t have any large projects underway.
If you’re a retail operation, don’t plan the move on Black Friday. Once you get to this point, you have an important decision to make.
Data Center Migration: To Forklift or Swing?
It is now time to pick your method for the data center migration. There are two commonly used options available in data center management: a forklift, or a swing data center migration. To “forklift” is to physically move all of the equipment from one location to the next. To “swing,” you’re essentially just duplicating the application hardware in a new environment.
When to do a ‘Swing’ for your Data Center Migration
Data center management should especially consider a swing maneuver if the current data center is: hosting production applications; more than 200 miles away from the new data center; unable to tolerate downtime; missing real time app failover abilities for production apps; or planning to upgrade a large portion of the equipment within the data center. Basically, the more chances you have for everything to fall apart like a very expensive jenga tower, the more you’re probably going to want to “swing.”
When to do a ‘Forklift’ for your Migration
Data center management should be fine to implement a forklift data center migration if the data center is: under 50 miles away from the new location; able to tolerate a full day of downtime; has sufficient failover capability; does not plan to refresh much of its equipment.
A Progressive or Immediate Forklift Migration
If the data center manager decides on a forklift maneuver, determine if the equipment should be moved in phases or in one sitting. This will largely depend on whether you prioritize minimal downtime, or a completed project. Phasing in the equipment to the new data center location can help reduce downtime, while a “big bang” data center migration will take less time overall. Do you want to rip the band-aid off and tear a few hairs out in the process or do a slow peel?
Outside Help Can Make Your Data Center Migration Go More Smoothly
You have two options for a data center migration. You can:
- Use your in-house data center management team exclusively.
- Hire outside help, such as an expert to spearhead your data center migration
If your team does not have enough redundancy within its roles, it may impede other processes within your organization. For instance, if unexpected demands arise from their primary role during the data center migration. It’s quite likely something critical is going to come up. A Lack of redundancy in your roles can cause the guy who usually oversees equipment checks is putting out fires with the new location.
Even in many cases where the team is extremely qualified, and there are some extra human resources, it can be very beneficial to hire a data center migration expert to manage the project for a smooth transition. Sure, the Avengers are great on their own, but where would they be without Nick Fury?
Track Every Step of the Data Center Migration Process
Put tags on every piece of equipment to clearly identify what is moving in the data center migration. Some warranties can be voided in a data center migration, so double check all of the existing agreements. Inform your vendors if you have any service contracts to coordinate the data center migration. Check for any unique licensing restrictions on equipment to run simultaneously in both locations while your data center migration is in progress.
With planning, outsourcing, tagging, audits, and effective logistics strategies, we’re all set to have an effective data center relocation.
What About Retiring your IT Assets and Recycling them?
What about if you are getting rid of some of your equipment? If you’ve planned ahead, then it is time to look into selling your IT equipment and upcycling, not just IT or E-waste recycling.