Edge compute (distributed IT) installations have become increasingly business critical. Deploying and operating IT at the edge of the network, however, comes with unique challenges. Solving them requires a departure from the traditional means of selecting, configuring, assembling, operating, and maintaining these systems. This paper describes a new, emerging model that involves an integrated ecosystem of cooperative partners, vendors, and end users. This ecosystem and the integrated micro data center solution it produces, help mitigate the unique challenges of edge applications.
In a time where everything, everywhere is being “digitized”, we continue to see technologies like IoT, machine learning, AI, robotics, virtual/augmented reality, and data/video analytics being deployed. Businesses need these technologies to maintain competitive advantages and to stay relevant in their industries. But with these technologies comes latency, bandwidth, autonomy, and regulatory/security requirements that a centralized data center architecture cannot support. As a result, an end-to-end hybrid architecture, consisting of (1) local edge data centers, (2) regional edge data centers, and (3) centralized / cloud data centers, has emerged. And this hybrid architecture, including the local edge, must be highly resilient.
In this paper, we focus on common problems found in edge environments, and what it takes to design, assemble, deploy, and manage a robust infrastructure, so that business-critical applications remain available and personnel at the edge sites can focus on their primary tasks. Specifically, we believe the answer lies in having an improved ecosystem of partners and tools throughout the lifecycle, and a fully integrated micro data center solution that is born from that ecosystem.
- The ecosystem of partners – Consider an organization with dozens or hundreds or even thousands of local edge sites (sometimes multiple sites within a single facility). How do I deploy the IT and support infrastructure in a standardized way, monitor the equipment, maintain the equipment across all those sites? Vendors, integrators, service providers, and the end user collaborating throughout all phases of the edge data center ensures the right equipment is deployed, the right management tools exist, and the right operations/maintenance protocols are established.
- The Integrated micro data center – This is the complete system that emerges from the ecosystem of partners. It includes the IT hardware/software (e.g. converged infrastructure), the physical infrastructure hardware, and software tools for monitoring/management of the edge environment. An effective solution is easy to deploy, easy to manage, ensures the IT is safe from physical threats, and remains resilient in the face of power disturbances and outages throughout its lifecycle.
For most companies with edge data centers, it is costly, inefficient, impractical, and unreliable to design, implement, and manage these sites without this integrated approach. According to World Wide Technology, a global IT solution provider, the ability to pre-configure technology platforms and devices before shipment to site:
- reduces field engineering costs by 25 to 40 percent
- increases order processing speed by 20 percent
- reduces maintenance costs by 7 percent
Challenges of edge sites
There have always been challenges related to selecting, configuring, deploying, and maintaining IT infrastructure at a given location. Those responsible for edge compute sites, however, often face the additional challenge of having to do this over multiple locations with little to no onsite staff. Working with the right ecosystem eliminates or mitigates the challenges described below.
Selecting and configuring infrastructure components can be complex. One mistake can propagate to hundreds of installations if not identified early on. Infrastructure parts must be selected that not only support the intended application, but that are also compatible with each other and fit each local site’s conditions. This complexity leads to common mistakes we have seen at the edge such as:
- UPS undersized or grossly oversized for the expected IT load
- not considering UPS redundancy or UPS bypass (allows UPS replacement without downtime)
- not enough output receptacles (or wrong type) on rack PDU / UPS
- not enough U space in rack or spare space for expansion
- forgetting to order accessory parts kits needed to rack and cable everything
- not considering space constraints in rolling equipment into place
Physical infrastructure vendor: Simplify deployment & operations through configurators, reference designs, resilient infrastructure, and management tools
The physical infrastructure vendor provides the equipment to ensure the IT remains secure and operational. This includes the physical enclosure (rack) for the IT equipment, the uninterruptible power supply (UPS) to provide battery backup for the IT equipment, rack power distribution (rack PDUs) to provide the needed outlets for the IT equipment, cooling, environmental monitoring such as temperature, humidity, & water sensors, physical security cameras & access control, and management software. The physical infrastructure vendor should:
- create rule-based configurator tools based on IT stack needs (see Figure 3)
- develop micro data center reference designs
- understand the criticality implications and risk of downtime
- provide offers that address appropriate level of environmental, physical security, and management
- develop management tools with public APIs, so data can be accessible by any partner
- Enable offsite integration and testing allowing products to be shipped configured without risking damage or warranty
- drive interoperability with other vendors hardware & software upstream and downstream of their hardware
Systems integrator: Add value through complete integration of IT hardware, software, and physical infrastructure
The systems integrator is a company that brings together a range of solutions needed to meet the business objective. They serve a key role in the ecosystem, as they are the coordinator of all the parties involved. The systems integrator should:
- translate business need to IT requirements
- have expertise on the industry and technologies they deploy
- help determine the best IT stack needed to support the business application
- have a broad understanding of IT hardware and software integration
- have alliances with IT and physical infrastructure vendors
- help determine the physical infrastructure needs for the IT, based on power capacity, dimensions, weight, plug-types, criticality, etc.
Managed service provider: Operate & maintain edge infrastructure through management tools and digital services
A managed service provider (MSP), focused on information technology (IT) and/or operations technology (OT), is a company that helps operate the sites, addressing the staffing challenge on-premise. This often includes remote monitoring, proactive maintenance, repair and upgrade services. This is where collaboration becomes key. The IT, OT, and physical infrastructure vendors need to offer management solutions with open APIs that the MSP can use or tie into. The MSP becomes the eyes and ears of the micro data centers. They implement digital services offered by vendors. Everyone in the ecosystem needs the ability to look at the exact same data at the exact same time. One partner may be troubleshooting the problem, the managed service provider may have to go and fix it, so it’s helpful for everyone to work from the same data set. Cloud-based management offered by vendors provides the visibility and connectivity across the ecosystem. This is discussed in more detail in the following section.
The product of the ecosystem
- Pre-configured and tested
- Includes necessary infrastructure to maintain IT resiliency
- Uses open APIs & cloud-based software management tools
In today’s hybrid data center architectures, the edge of the network requires the same high resiliency as centralized or cloud data centers. But they are very different in two main ways: (1) they generally lack trained, onsite staff; and (2) there are multiple, distributed sites. This leads to some unique challenges that require us to think holistically and collaboratively when deploying IT.