- Date published:
- Author:Brian Wood
Great article from analyst Glenn Ford from 451 Research talking about the real impact of a client moving data centers.
Living so much in the virtual world as we do, it’s often easy to forget that a physical swap-out is far harder than just “Control Z”, reposition cursor, “Control V”.
The considerations below are important to keep in mind when selecting a data center to begin with.
As with most things in life, you never get more than what you pay for — and sometimes you don’t even get that.
Emphasis in red added by me.
Brian Wood, VP Marketing
MTDC Customers Tend to Be Sticky
17 Dec, 2012
Why do you want to move your datacenter? Frequently, companies move their datacenter to gain expansion space, to reduce the cost of maintaining internal IT infrastructure or as part of a server consolidation effort to reduce their overall footprint and associated costs. However, really, does it make sense to move a datacenter?
It may makes sense if you have aging internal IT infrastructure that you are replacing/upgrading or if you need a significant datacenter expansion to support current and future growth, but does it make sense if you are colocating your infrastructure in a multi-tenant datacenter (MTDC)? Probably not, and this report will look into why customers tend to stay once they are in a MTDC.
Planning a move
Moving a datacenter is a very difficult and complicated task. You are physically moving pieces of highly sensitive and expensive electronic equipment that contain data and processes critical to your business. Datacenter moves involve identifying and analyzing applications and mapping dependencies to determine the order for bringing the systems back online, configuring and reconfiguring network connections, dealing with data storage concerns, and many other considerations.
An analysis of the costs, benefits and risks of moving your datacenter must be considered. Areas to explore are the costs related to the impact on business operations, insurance for the move, consultants to assist in planning and execution, movers, additional/new equipment and other related costs. These would be compared with the savings, if any, and the benefits to the business to remain where you currently maintain your datacenter. The numbers rarely work out in favor of a datacenter move.
Different methods for moving
The typical window to move a production datacenter is 24 to 72 hours, but it can take months of planning and preparation and for larger moves, this may be done in phases over months. The list below contains a few methods we have seen or heard of companies using to move a datacenter.
- Use new equipment at the new site including racks, servers, network gear and data storage. After installing the new equipment at the new site, restore the systems and data from DR/backup systems, perform tests to verify all systems are functioning properly then schedule the actual cutover to the new site. This method is one of the safest to implement but one of the most expensive.
- Shut down all systems and relocate the racks, servers, network gear and data systems to the new site. Reassemble the infrastructure, test it out and activate the production systems. This method has significant downtime associated with it and usually is not used, although some smaller businesses have moved datacenters using this method or this method is sometimes used to move a DR site.
- Businesses with redundant systems have ported the failover systems to the new site and established all systems were working properly, cutover operations to the site. All remaining systems are then moved and reconnected in a failover configuration. This method presents high risk to system operations in the case of a failure and requires the core-supporting infrastructure to be in place.
- Businesses with virtualized systems may not have as many considerations to address and will typically use variations of the options listed above. Most companies have a few legacy systems that cannot be virtualized that will require extra planning and possibly extra costs to move.
The options above are only a few of the combinations that have been used to move a datacenter. Most of the options employed have the common elements of some new equipment (usually network gear, racks and PDUs) and have costs for transport, wiring, networking setup/testing, security and business continuity insurance. New equipment is usually needed because businesses do not have enough surplus equipment to support the move. A move is also a great opportunity to replace aging supporting infrastructure with new equipment.
The costs for preparing a new datacenter site include setting up the supporting infrastructure. At a high level this includes building out the floor space or installing cages, installing connectivity and power drops, and running fiber to backup systems. Equipment racks for servers and network systems need to be set up, and fiber, wiring and power run to the racks. Network routers need to be configured and tested; don’t forget the load balancers. Network switches and PDUs need to be installed in racks. The setup and testing of this infrastructure will require contract labor.
You may wonder why you need to contract labor to perform the initial buildout. Typically, IT support staff are right-sized or under-staffed and are fully allocated in support of current operations and do not have the time available to set up infrastructure in a new datacenter. Your options are to contract out the setup activities or hire supplemental staff freeing your team to perform the setup. A common option businesses employ is to contract out the setup of everything but the network equipment and have internal staff perform those activities once the heavy lifting is done.
In addition to contract labor costs already mentioned, you will need to hire a moving company that specializes in moving sensitive electronic equipment. Insurance is also recommended in case an unplanned event occurs during the move that will affect business operations. Events like traffic accidents during moving that can damage critical equipment. Moving older legacy systems can be very tricky and bouncing another in the back of a truck can sometimes cause hard drives and other components to fail. Older systems on occasion fail to start up after being turned off and allowed to cool down.
Splitting your systems up into different trucks and having them take different routes can manage some of this risk. Police or escort companies are frequently hired to reduce traffic issues and to provide security during transit.
To some extent, insurance can offset the economic impact of events that cause delays in getting critical production systems back up, but there is always some reputational impact that cannot be easily repaired. Customers tend to remember that systems/services important to them were not available.
We have heard from MTDC providers that we have talked to and we have seen in our MTDC customer win roundup reports that customers place a high value on customer service and high-quality managed services. If MTDC providers are building customer satisfaction and relationships in these areas, it is likely that they will keep existing customers and attract new ones.
Managed services are becoming more important to customers because competition for IT talent is growing. Businesses want to focus on their core competencies and not have to worry about finding, training and retaining talent to support critical IT systems. They are looking to providers for managed services to fill the gaps where they need support.
MTDC operators who do not have a managed services offering are partnering with companies for specialized skills to support their customers. A few providers feel that they need to focus their core competencies of running datacenters and leverage partnerships with others to support customers seeking additional support services.
What makes MTDC customers sticky is the cost and the risks of moving out of a facility into another. One factor that customers consider is the longevity of the relationship with a MTDC provider that will avoid a costly move in the future. Contributing factors to the selection process include room for expansion, ability to scale bandwidth to meet a growing demand, a managed services offering and customer service that will meet current and future support requirements.
Another factor that may complicate a move for a customer is tightly coupled application systems. In a perfect scenario, programs and configurations would use hostnames and let DNS provide the IP addresses; however, programmers do make mistakes and include hardcoded IP addresses and other connection information in application programs. This will cause issues when moving systems to a different network space. Network engineers can sometimes use tricks like network address translation to solve the problem but they do not always work. The address space your systems have hardcoded may overlap with your MTDC provider’s network need for access to data backup systems. This problem is usually one of those operational hiccups discovered at the new location and can delay the move until a solution can be implemented.
A key consideration that customers evaluate before moving their datacenter is the cost of the move and the overall impact to the business. Of course a small footprint of a handful of cabinets is much easier to move than a cage of 50-100 cabinets. We are aware of datacenter moves for financial firms with only 5-10,000 square feet of datacenter space that have cost millions of dollars when everything is factored in.
We have said it many times; moving a datacenter is a very expensive and difficult project that can take months to plan and execute. Most moves are successful but all have a few unforeseen bumps along the way.
The 451 Take
We believe that customers will continue to stay with their current MTDC provider as long as they are getting the level of support and service they expect. The costs and complexities associated with a datacenter move are not worth the risk to business operations and corporate reputation. MTDC providers are continuing to improve the cost-effectiveness of datacenters in order to remain competitive and are moving up the stack to provide virtual and cloud services that can reduce the number of physical servers needed by customers to support their business.
We believe that the education, managed services and consulting services MTDC providers are adding as they move up the stack will help to build strong customer relationships and keep customers in place. MTDC providers are educating customers on the importance of having a disaster-recovery/business-continuity (DR/BC) plan in place and are assisting them in the development and implementation a DR/BC program. MTDC operators are offering DR services that provide office space to support a recovery event. A few providers have arranged with local hotels to provide living accommodations for the recovery teams and their families. We believe colocation and wholesale datacenter customers will continue to be sticky.