The global transition into a remote workforce
has caused an even higher demand for cloud and data center resources. Data
center usage will continue growing for a variety of reasons, but generally, server
management is easier (and often less costly) than on-prem management. If they
haven’t already, many companies are now seeking out how to leverage the
efficient resources offered by data centers.
Researching data centers is no easy task;
there are a load of aspects to consider. Whatever caused the search to begin,
look no further. We’ve compiled a list of the most important features to look
for and the questions to ask to make sure you’re getting the most out of your
This information will assist in your search but keep in mind that selecting a data center starts with the unique needs of your business. Work with the appropriate professionals to determine your specific technical requirements, then record any legal or compliance obligations, and finally compare facilities that check those boxes. It doesn’t matter what a data center offers if they do not meet your business requirements.
Below is a write-up of what questions to ask, what each means, and its recommended answer to help you gain an understanding of all the most important aspects of a data center. For a quick, fillable checklist, download this version which includes additional sections regarding compliance, sustainability, corporate information, and cost.
- Business Requirements
- Redundancy and Resiliency
- Service and Support
- Compliance (download only)
- Cost (download only)
- Sustainability (download only)
- Corporate Information (download only)
First, record your custom requirements for any of the categories listed above and any legal, regulatory, statutory, or contractual requirements you may have. A data center should be able to provide documentation as evidence of its compliance with appropriate standards. Read our certification guide for an explanation of different compliance frameworks and certifications.
The Uptime Institute’s 2020 global survey of IT and Data Center Managers found that on-site data center power problems were the primary cause (over one-third) of major outages, 75% of which respondents said were entirely preventable. Unplanned downtime leads to financial loss (among other things), so it’s important to be confident that your provider can keep business running and not negatively affect yours.
What is the total power capacity of the data center?
This is dependent on the amount of power you need. Determining your unique resource requirements is an intimidating and difficult process but can be handled by an IT professional. They will take an inventory of all hardware resources and perform load tests to determine your capacity. It would be helpful to know how your power usage has increased year over year to give you an idea of how much power you may need in the years to come. Generally, small data centers average a power density of up to 15 MW (Megawatts), medium range from 15 MW – 40 MW, and large above 40 MW.
What is the power density per rack?
To avoid waste, it’s important to select a power density above but close to your requirements that would accommodate potential growth. Underestimating power needs can result in future disruptions when increasing capacity. In general, the average in 2020 was 8.4 kilowatts per rack. Oasis runs higher (up to 30 kW per rack) because we require it for heavy processing workloads.
How is power fed into the data center?
When it comes to power, you want all sources to be as independent as possible. This way, if there is a power failure, only one portion is affected. Look for independent substations with dedicated feeds and diverse routes into the data center. A data center located on a separate power grid from your organization is also ideal.
What is the PUE (power usage effectiveness) score or DCIE (data center infrastructure efficiency) percentage?
PUE is the ratio of the total amount of power used to the total power delivered to computing equipment. The closer it is to 1.0, the better. DCIE is the inverse of PUE. It is the amount of energy utilized by the computing equipment to the total power available. A DCIE of 67% or higher is suggested.
What’s the facility’s uptime?
Uptime is the term used to describe system reliability or how often the system is up and running (as opposed to downtime). An uptime of 99.999% (or “five nines”) equates to 5 minutes and 16 seconds of downtime per year. The less downtime, the better. Keep in mind that the decimals add up—99% doesn’t sound significantly different but is, equating to nearly 88 hours of downtime—could your business afford that much?
The building itself—where, why, and how it was constructed is important primarily for physical security and efficiency. The way a data center was built determines how well it will be able to protect and maintain equipment.
Was this space retrofitted or purpose-built?
Choose a data center that was built specifically for the purpose of housing and protecting servers. Data centers have extremely specific infrastructure, power, and efficiency requirements that a retrofitted data center will struggle to meet. Other features of a purpose-built facility will be thick concrete or steel walls and roof, plus minimal doors and windows for the highest structural integrity.
Is this a high-risk area?
Find a data center located outside of moderate and high-risk zones to avoid hurricanes, tornadoes, tsunamis, and other natural disasters. If the facility is in an area of high wind, make sure the walls were built to withstand winds of 215 mph.
What public facilities are nearby?
A data center located outside a one-mile radius of public facilities, such as a major highway or airport, is ideal for visiting convenience, yet far away enough for safety.
What is the primary method of cooling?
When hardware works, it creates a large amount of heat. That heat needs to be replaced with cool air in order to keep the hardware running properly. Cooling is one of the most critical features of a data center and another reason to avoid retrofitted facilities. We suggest a hot aisle/cold aisle design (avoid inefficient raised floor systems), which is the most effective design for zoned climate control. The data center should also be able to provide maintenance and replacement records for their HVAC system.
The importance of proper security cannot be stressed enough, especially with the frequency and sophistication of attacks increasing year after year. A good data center (and its affiliated service providers) will provide layered security controls using a defense in depth approach to reinforce the safety of your assets.
What layers of security do you have in place?
This is not an exclusive list, but here are some good physical security features to look for in a potential data center: perimeter fencing, armed guards, keycard-zoned areas, biometric verification, authorized visitor list, locked cages, 24/7 surveillance with archived footage, ID badges, man traps, and fire prevention systems. Most data centers include a manned SOC/NOC (security operations center/network operations center) to monitor the facility’s infrastructure.
What staff is on site and when?
The facility should be staffed 24/7/365 inside, outside, and at all potential entry points (though uncommon, ours has 24/7 armed guards and a dedicated fire department). Ideally, the staff would be directly employed by the data center to provide dedicated, facility-specific knowledge.
What are the visitor guidelines?
You need to let the right people in and keep the wrong people out. Features that help with that include keycards, an authorized visitor list, visitor escorts, biometric verification, access and activity logs, and around-the-clock monitored surveillance.
simply, connectivity is a data center’s ability to communicate with the outside
Is it carrier-neutral?
A carrier-neutral data center allows the highest level of flexibility for its tenants. It can connect with a variety of providers, allowing not only a choice for tenants but also plenty of scalability. Having more connections spreads the traffic out, avoiding congestion. Some data centers will also allow you the option of your very own fiber optic cable, providing a direct connection between the facility and your business.
Is there high-quality blended internet service?
This option allows you to blend multiple internet services to ensure the fastest speed and the highest reliability. If one internet service provider goes down, you won’t. The more diverse carriers they can offer, the more resilient and cost-effective options are available.
Are DDoS mitigation services available?
A Distributed Denial-of-Service (DDoS) Attack is when an attacker floods a server with internet traffic to prevent users from accessing connected platforms. Mitigation services can keep users online during an attack and then prevent another from happening. Some data centers have perimeter-based DDoS protection built into their contracts.
Redundancy and Resiliency
Redundancy and resiliency are closely linked together. Resiliency considers a data center’s planning, management, and prevention methods (all of which should be documented and regularly tested) in reference to its ability to handle a component failure. Redundancy refers to its backup equipment capacity, which contributes to the facility’s overall resiliency. A “fully redundant system” or “2N+1” means all critical components are backed up. Both redundancy and resiliency are telling factors in the way a data center maintains business continuity.
Are there dual/diverse paths and entry points for connectivity purposes?
The main goal with diverse routing is to ensure that independent lines do not share the same path or underground ducting/cabinets. Without any shared entry points, you can ensure the availability of your connection in the event of a disruption.
What backup sources of power are available?
All data centers should have diesel generators ready in a secure external location in the event of a power failure. Additionally, some data centers will have this important feature: uninterruptible power supplies (UPS), which hold power before the generators kick in so that there is no drop while the system transitions to them. If a data center doesn’t have integrated UPS, they can still be added to completely avoid downtime in the event of a power failure (remember: power manages all infrastructure, lighting, AC, etc).
What is its disaster recovery plan?
This should include a schedule of testing for redundancy and equipment reliability, and the results of any tests that have been performed.
These may not be your first priorities but are likely to make working with and in the data center much more convenient.
What is the media handling process?
There should be secure, managed shipping and receiving with a fully documented chain of custody procedure implemented from receipt to release/handover.
Is a remote hands service included?
It’s helpful to have a 24/7 remote hands service available to help with cabling, technical support, etc. Find out what’s included—some offer much more than others. Oasis’ Advanced Remote Hands, for example, provides racking, inspection, testing, installation, and a variety of other helpful services to have at the ready.
Do they offer…?
- Technical staff available for support, including
- Access to additional provider services like IT
and managed cloud?
- Rack and stack
- Power cycling
- Staging area available
- Secure storage or lockers
- Meeting space
Your data center is an integral part of your business. It’s necessary to know if it can
meet your needs today and as your organization grows. With
the right partner, you get much more than servers on a rack—you get value-added
service that enables your growth and adds to your bottom line. If you’re
not benefitting from your data center, it’s time to change.
To learn more about our high-performance data centers, contact us at firstname.lastname@example.org or fill out our simple contact form.