What are Micro Data Centers?

Definition

A micro data center (MDC) is a small-scale modular data center that includes all the compute, storage, networking, power, cooling, and other infrastructure required for a given workload.

Overview

In recent years, applications including IoT (Internet of Things), content delivery, and 5G have created a large demand for low-latency access to data processing and data storage. Traditional centralized data centers, such as those used by AWS (Amazon Web Services) and Microsoft Azure, weren’t designed with those use cases in mind. While these large-scale data centers create economies of scale and enable cloud computing, they cannot economically bring data processing close enough to end-users for these distributed workload use cases.

Ranging in size from a single 19-inch rack to a 40-foot shipping container, micro data centers are small enough to be deployed in locations where a traditional data center would be impractical. This allows organizations to solve the low-latency data processing and storage challenge in a cost-effective way.

While there is no single standard for micro data center design, an MDC must be able to house the full technology stack an application requires. As a result, these components are common in MDCs:

  • Server racks/enclosures
  • High-performance servers
  • Networking equipment
  • Power Distribution Units (PDUs)
  • Uninterruptable Power Supplies (UPSes)
  • In-rack/in-container cooling
  • Environmental monitoring
  • Physical security measures
  • Fire suppression technology

Generally, MDCs are modular by design. This means components can easily be swapped if there is a failure or requirements change along the way.

How micro data centers relate to edge computing

Often, the terms “micro data center” and “edge computing” are used in similar contexts. While they are related, they’re not the same thing.  What’s the difference? Edge computing is a type of distributed architecture while MDCs are a type of data center design.

For practical purposes, think of it this way: MDCs are a means to help implement edge computing in the real world. Edge computing is the “what” while micro data centers, along with a network backbone, are part of the “how”.

Benefits of micro data centers

By bringing data processing closer to end-users and enabling organizations to distribute their data center footprint, MDCs deliver several business benefits including:

  • Low latency: This is the fundamental issue MDCs help address. By bringing the “client” and “server” computers geographically closer, MDCs can lower latency significantly.
  • Increased resilience: A failure, such as a cut MPLS (Multiprotocol Label Switching) line, that leads to a traditional data center becoming inaccessible can bring production to a halt. Even if fault tolerance is part of the design, it’s often N+1 or 2N at best. With MDCs, having dozens of failover options becomes feasible.
  • Faster deployment times: Often, micro data centers can be shipped partially or fully assembled. Additionally, because their small size makes accessing adequate space and power easier, MDC provisioning is significantly faster than a traditional data center build-out.
  • Standardization: By enabling a standardized and repeatable design, MDCs can reduce opex and maintenance costs.
  • Scalability: Micro data centers allow businesses to incrementally scale up based upon demand. This allows for a more elastic and agile approach to infrastructure that is conducive to rapid growth.
  • Reduced costs: Traditional data centers and server closets come with high upfront capex and ongoing opex. The standardized and modular design of MDCs lower opex, while the small footprint can help keep capex low.

How Micro Data Centers Work

The nuts and bolts of how any given MDC works will depend on the specific components (switch gear, servers, power, cooling) installed. However, at a high-level most micro data centers are deployed, provisioned, and maintained in a similar fashion.

Often, micro data centers ship preassembled, meaning the equipment is preinstalled in a server rack(s). Upon receipt, IT staff will install any additional components and connect network and power cabling within the MDC. From there, IT can connect the MDC to an existing utility power source or have an electrician provision a new circuit for larger installations.

Once the MDC is up and running, IT staff or an MSP (managed service provider) are responsible for configuration and monitoring of the equipment. For large replicated MDC deployments, configuration management tools (e.g. Ansible) and enterprise network monitoring solutions (e.g. SolarWinds) are common.

Examples of Micro Data Centers

Micro data centers can be helpful almost anywhere high-density, small-footprint compute resources are in demand. However, in some of the applications, what differentiates an MDC from a traditional IDF (intermediate distribution frame) or server closet deployment isn’t well defined. Therefore, we’ll focus on the edge applications for MDCs here.

Popular micro data center use cases today include:

  • IoT: The devices that make up the Internet of Things generate huge amounts of data and enable new automated workflows. In order to get the most out of IoT, enterprises need to reduce latency and shift storage and compute resources closer to the “things”. For example, for a delivery drone or smart car, a few hundred milliseconds of latency can make a world of difference.
  • Content delivery: High-definition video streaming is one of the most intuitive use cases for a CDN (content delivery network). But any application where static content (images, PDFs, web pages) need to be delivered and performance is a concern is a potential use case for a CDN.
  • 5G: There is plenty of hype around 5G, and for good reason. The high throughput cellular technology has plenty of potential. However, the millimeter waves and microwaves used by 5G technology don’t travel far. This means many small 5G stations are required for sufficient coverage of a given area. MDCs will be essential to building the underlying infrastructure that enables end-to-end 5G connectivity.
  • Hybrid cloud architectures: For applications using machine learning and artificial intelligence, the latency and bandwidth costs associated with public cloud platforms can become a problem. Micro data centers enable business to get the benefits of the cloud while also bringing compute and storage resources on-premises where it makes sense.

Key takeaways

  • Micro data centers are small-footprint data centers.
  • Edge computing is a “what” while micro data centers are a “how”.
  • Micro data centers bring compute and storage resources closer to end-users.
  • Micro data centers can reduce costs and deployment times while increasing resilience and scalability.