This content is part of the Essential Guide: Emerging data center workloads drive new infrastructure demands

How an IoT deployment reshapes network and storage management

Data center admins need to know the unique network and storage requirements that come with an IoT deployment, and evolve their IT management practices accordingly.

The internet of things can offer organizations up-to-the-minute business insights gleaned from an ocean of data, typically collected from sensors. But those sensors require proper architecture, deployment, management and support, as well as unique storage and retention demands. This means IT planners and architects must evaluate the implications of the internet of things before embarking on an initiative.

Here are some important questions for data center teams to ponder before an internet of things (IoT) deployment.

Is your network suited for an IoT deployment?

Each IoT device produces data that must traverse the network to a point where data can be stored and processed, and every IoT sensor or device multiplies IoT traffic. One of the first barriers to an IoT deployment is network bandwidth limitations that can result in bottlenecks and performance degradation.

This can be a complex issue. Network suitability will depend on where the IoT devices are physically located, the network's current bandwidth capabilities and the way in which the network is architected. For example, if a large collection of IoT devices are placed remotely, latency in the internet link between the data center and the remote site may occur. If the devices are placed locally, latency may still occur on the network segments where those devices reside. To understand how IoT devices impact network performance, establish test projects, monitor traffic patterns and measure network performance.

For local IoT networks, upgrade slower segments to enable faster bandwidth, deploy multiple links to boost bandwidth or rearchitect the network to reduce IoT traffic contention. For remote IoT deployments, consider edge computing to store and transform IoT data -- then move only the preprocessed data back to the data center for use.

How should you manage an IoT deployment?

Beyond the physical requirements of installation, maintenance and replacement, IT teams need to configure, test and manage every IoT device. Ideally, an IoT deployment is manageable from a remote administrator's desk, just like most current server, storage and network devices.

IoT device management normally involves a software tool from the IoT vendor, but administrators must consider how the software integrates with current management tools, such as System Center Operations Manager. If a vendor offers a plug-in that adds IoT support to a major management platform, it can minimize or even eliminate integration problems.

Deploying and maintaing IoT devices

IoT installation can be as simple as looping devices over clothes hangers or inserting them into mass merchandise. It could also be complex, such as installing new intelligent valves, solenoids and other instrumentation into critical machinery. Once installed, you must configure and test the devices, which may need periodic calibration. Many IoT devices rely on battery power, requiring periodic maintenance to replace and test batteries. And eventually, those IoT devices must be removed, maybe recycled and perhaps replaced with other devices.

IoT device management can affect IT workflows, especially when workflows are formalized in orchestration and automation systems. Administrators may need to orchestrate IT workflows to handle the initial configuration and testing of new IoT devices. Additionally, they may need to update existing workflows to add steps and policies that address IoT devices. Workflow changes can be cumbersome and require revalidation and approval from senior management.

How much storage do you need for IoT data?

First, consider where you should store IoT data. Traditionally, data is passed over the network and stored in the main data center where a business can exercise maximum control and security. But this also imposes a significant capital expense in infrastructure, and may lead to network bottlenecks. Alternatively, the business might move data to an outsourcing facility located closer to the IoT devices, reducing latency while maintaining a centralized infrastructure. However, more businesses are opting to store raw IoT data at the edge -- placing compute and storage assets as close as possible to where IoT devices are deployed, and then passing only processed or transformed data back to the main data center. This can alleviate some potential network performance issues, but complicate the overall infrastructure. A more dynamic and flexible storage location for IoT data is the public cloud, which eliminates capital storage expenses.

Next, consider how much storage you'll need. It might seem like a mundane capacity planning issue, but IoT data volumes and retention periods can create erratic storage capacity and utilization patterns. It's impossible to place a definitive number on IoT capacity because it depends on how much data you collect and how often you use that data. For example, if you're only collecting data for 24 hours before performing an analysis, it will require far less storage capacity than a 30- or 90-day data collection and analytics project.

Finally, determine your storage performance requirements for IoT data, such as online, nearline or archival. For example, a data center may use fibre channel or solid-state disks for top performance, SAS disks for moderate performance and SATA disks for archive-grade storage performance.

How long should you retain IoT data?

Data retention concerns are certainly not new, but IoT adds a new dimension to the retention problem for IT professionals. There are few real IoT data retention standards yet, but retention will always impact storage capacity requirements.

Generally, data retention for an IoT deployment depends on the data being collected and how it's used. For example, if an organization connects IoT data from sensors around a commercial aircraft or industrial installation, that data will likely face a long retention period to accommodate long-term modeling, analytics and potential forensics. But if you're using IoT sensors to gauge retail shopper traffic around holiday displays, that data will only be valuable for short-term decision making and can be discarded more quickly. This notion of short-term data collection and usage can be unsettling for some IT and business leaders that have been shell-shocked by compliance and retention regulations.

Include IoT data in your organization's formal data retention and destruction policies, and treat it as a new class of data with unique retention guidelines.

Next Steps

What exactly is the internet of things?

IoT architecture in the data center

How to strengthen your IoT infrastructure

Dig Deeper on Emerging IT workload types