When it comes to data centers, one size does not fit all. Data center needs vary depending on their structure, physical limitations, density requirements and more. Here are four common data center types including onsite, colocation facilities, hyperscale, and edge data centers, as well as their use cases and industry trends.
Onsite Data Centers
Onsite data centers, also referred to as enterprise data centers are housed within a company site/ campus. They can be changed or expanded as the company needs and are easily accessible for general maintenance or troubleshooting. Onsite data centers increase performance because the equipment is stored at the company location. This typically keeps data secure, and the data can be controlled by the company that owns it.
According to Bold Business, many companies choose alternatives to onsite data centers, because they can be expensive and take up a lot of resources. An onsite data center needs a reliable power supply and cooling system, a large network, a security system and more, all of which can be hard to support in-house.
Colocation facilities rent or lease space to multiple organizations to house their data centers. Many organizations utilize a colocation facility, commonly referred to as a colo, for their equipment to benefit from their cooling, physical security, up-time and monitoring guarantees, maintenance and locations.
There are two types of colocation facilities: retail and wholesale. Retail colocation facilities have smaller spaces for rent and only provide infrastructure equipment, such as the rack, for rent. These facilities also maintain the switchgear, generators, cooling, UPS (Uninterruptible Power Supply) and other equipment that keeps the location running. Wholesale colocation facilities rent out the entire data center unit and provide larger spaces. These facilities are typically better for companies that need a lot of space and may expand in the future. Both types of colocation facilities lease power as a cost per kilowatt.
Why use a colocation facility? In his article, Rowland Kinch shares insight into why. If an organization does not have 24-hour maintenance and/or an IT team to deal with IT issues, colocation sites will provide that service. A guaranteed internet connection is also available at a colocation site, which may not be available at the company’s site. Lastly, the maintenance and upkeep of data center facilities can be expensive, and colocation facilities provide these services, as well as security.
Trends in cloud computing keep changing, and to keep up with them, colocation facilities need to change as well. As the cloud becomes more popular, according to Yevgeniy Sverdlik in his article “Four Data Center Colocation Trends to Watch in 2019”, colocation facilities are making changes to their digital tools to feel more like the cloud. These changes have been made to the user experience and the functionality of the data center. A few of the main components of these changes include Sverdlik states, automation, APIs (Application Program Interfaces) and the combined management of the various types of data centers.
To summarize, Steve Smith, Chief Revenue Officer at Coresite, as quoted in Sverdlik’s article, states that “as enterprises become more sophisticated in leveraging multi-cloud platforms, demands will increase around API functionality, capacity management, and service provider availability on those platforms”.
Hyperscale Data Centers Hyperscale data centers are large and can house thousands or millions of data center servers. These data centers are meant to be scaled, which can be done by increasing the power of the system or adding more data center racks and equipment. Companies with large amounts of data benefit from the scalability of hyperscale data centers.
Hyperscale data centers are a popular option, and with the increase in data storage needs, these data centers will only continue to expand in size. In his article, “The Eight Trends That Will Shape the Data Center Industry in 2019”, Rich Miller quotes Compass Datacenter’s CEO, Chris Crosby, about the future of hyperscale data centers. Crosby predicts that the hyperscale sector will produce about the same amount of capacity for data centers in the next two years as it produced in the previous five years. That is a bold prediction, but like the cloud, artificial intelligence, machine learning and the Industrial Internet of Things are all rapidly growing, and more storage will be needed to hold the data.
Edge Data Centers Finding a definition for edge data centers is not easy. Edge data centers are similar to their traditional counterparts, except that they are smaller and built on the edge of the end user. Edge data centers decrease latency, increase capacity and improve connectivity.
The Telecommunications Industry Association (TIA) created a great chart (below) that helps define edge data centers by comparing them to traditional data centers.
|Category||Traditional Data Center||Edge Data Center (EDC)||Comments|
|Location||Site selection is well planned and is strategically located||Anywhere, downtown to remote, not readily available, hazardous; latency may force a compromise||A large multi-story DC vs an EDC which could be a 1/2 rack in a cabinet|
|Building Types||Usually newly purpose built DC facility with fully integrated systems & support; potentially multi-tenant facility||Many options: 1) modified cell tower shelter; 2) modified cabinet; 3) Drop and Plug Shelter; 4) Drop and Plug Cabinet; 5) IDF closet in a Building; 6) Co-located in a Central Office or Data Center; 7) In box on a light pole, etc.||There are many more possibilities for EDCs and most of these will require site preparation. Scale-out will favor standardized drop & plug designs.|
|Proximity to end-users or connected devices||Could be very far, even cities or counties away||Typically, very close, determined by latency requirements||Latency is a key driver for EDC locations.|
|Application||General purposes; could support thousands of applications||Few; specific and latency sensitive applications||EDCs provide smaller cloud capacity to support varied applications.|
|Latency||Higher, due to distance from end-user||Specific latency requirements||RTT <20ms may require EDC. EDCs can address specific business needs vs more generic, centralized DCs.|
|Availability||Mission-critical data, applications, and services per ANSI/TIA942 standards||Smaller scale, still mission-critical, with reduced time added to data, applications, and services. Modified ANSI/ TIA-942 standards.||Critical systems, power, cooling, network capacities and others will need to be balanced against scale-out requirements.|
|Size||Varies by design, typically larger than 100s, or 100,000s sq. ft.||Smaller, perhaps a few racks||Think small for EDCs - For every 1 DC there could be 100s or even 1,000s EDCs|
You can find edge data centers in many locations. For example, one could be found near 5g technology, by a cell tower, helping in a disaster recovery site or even behind a commonplace of business such as a gas station. Edge data centers sound like onsite data centers, but an offsite company or colocation typically manages them.
Edge data centers are continuing to trend in the market as organizations look to be more efficient by building their data centers closer to their location to reduce latency.
Data Center Technology
What kind of data center should I use? What are the requirements for my data center? These are common questions when trying to determine the best data center design, and with trends always evolving it may seem difficult to choose, but with the right information, the choice will be clear.
Are you looking for custom data center products without the custom lead time?
DAMAC, a division of Maysteel, manufactures custom and standard data center products including aisle containment systems, server racks and cabinets, modular systems/structures and more. Learn more
©2020 Maysteel Industries, LLC
Stay up to date with all Maysteel manufacturing news, updates, and trends. If you don't love it, unsubscribe with just a click.