Manage Learn to apply best practices and optimize your operations.

What 2019's 'summer of outages' means for IoT

In the summer of 2019, several cloud service providers experienced nagging bouts of unplanned downtime, impacting thousands of businesses. Google had an outage in June, which brought down several of its most popular services including Search, Nest, YouTube and Gmail, and was hit by another major outage in early July. Apple also experienced a widespread cloud outage in July, which affected the App Store, Apple Music and Apple TV. Cloudflare, Facebook and Twitter also had problems.

The macro conditions underlying these outages can often be boiled down to increased internet complexity or rushed-to-market software releases. While IoT was not involved in these recent problems, the implications of such unpredictability are significant for any cloud reliant IoT project, especially those with lives and safety depending on them.

This is because IoT and cloud computing are increasingly intertwined and symbiotic technologies. IoT devices generate huge amounts of data, with the cloud often serving as the central data collection and analysis repository. For example, consider a large multinational enterprise with IoT-connected thermometers across hundreds of factories, each one constantly generating data for analysis. These thermometers might be connected to other IoT devices and services, such as a factory manager’s remote, smartphone-based thermostat app. All of this requires superior speed and availability to work, making the recent spate of outages a major cause for concern.

Industrial IoT applications like this are just the tip of the iceberg. It’s one thing for factory thermometers to go down due to the cloud, but what happens when IoT is managing something even more critical, such as hospital systems and equipment?

The recent outages shouldn’t dissuade IoT projects from leveraging the cloud because, in many cases, the cloud offers higher levels of security, reliability and delivery speed than organizations can deliver themselves. But it does mean these organizations must be discerning and proactive about protecting themselves, especially if their IoT applications are mission critical.

Monitor the cloud yourself

Assurances from a cloud provider regarding availability and speed, round-trip time of packets traveling to and from your connected IoT devices, can give some peace of mind and a sense of the provider’s overall infrastructure health. However, this should be considered supplemental information only and cannot be relied upon exclusively for ensuring that IoT device connections are reliable and fast.

This type of direct to the cloud and back type of monitoring is not necessarily indicative of reality. Cloud service providers have partnerships with internet service providers (ISPs) and better network intelligence on how to route traffic. This means that, whenever possible, cloud monitoring will bypass the broader internet infrastructure that IoT device data must traverse, keeping packets in transit on their own networks and optimizing speed from point A to point B, and vice versa. This can result in a skewed, overly positive sense of IoT device connectivity and communication speed, because in the real world, networks and other external elements can get in the way.

Don’t track the cloud from only the cloud

While it’s critical for organizations with cloud-dependent IoT projects to do their own monitoring, they should never monitor the cloud from cloud-based infrastructure only. For the reasons outlined above, you might get a warped view of actual performance. Never monitor the cloud from the same cloud provider that’s handling your IoT project. If this cloud goes down, you’ll be blind to how your co-located IoT system is doing. You must make certain your monitoring vantage points are a mix of backbone, ISP, wireless and other node types.

Monitor IoT device availability

In an IoT world, devices essentially are the end users, so it is important to consistently monitor them and ensure they are reliable and interoperating with other IoT devices with exceptional speed. Since cloud service providers’ infrastructure consists of datacenters and other servers spread across the globe, a problem can occur anywhere and impact isolated segments of IoT devices.

That’s why it is critical to have as many eyes as possible, in all the key geographies where you have IoT devices running, as well as from the various network vantage points through which your IoT devices connect to the internet. This will put you in the best possible position to proactively detect IoT outages or slowdowns. Combining this with deep analytics will give you a head start in addressing the problem, whether it’s related to the cloud or not.

Have redundancy plans in place

If your IoT project supports a mission-critical process, you should consider having a multi-cloud strategy as a form of backup and protection. This might require a good amount of work, but it’s often worth it. You’ll need to make sure all the key phases of an IoT project, namely real-time data processing and storage, can be quickly ported over to another cloud in the event of primary cloud failure. This means testing failover strategies in advance to ensure cloud-to-cloud interactions are fast and reliable enough to support real-time data replication.

Final thoughts

The cloud has many attributes that make it ideal for supporting IoT projects. Not surprisingly, growth in IoT data has led to cloud service provider growth and expansion, which supports more IoT data and projects. Together, the cloud and IoT represent a set of inextricably linked technologies of the future.

But organizations running IoT in the cloud must proceed with caution. If we’ve learned anything, it’s that even the strongest businesses in the cloud industry can — and inevitably will — go down. It’s up to you to take the steps needed to better prevent your IoT project from going down with them. In fact, this is something you can’t afford not to do.

All IoT Agenda network contributors are responsible for the content and accuracy of their posts. Opinions are of the writers and do not necessarily convey the thoughts of IoT Agenda.

Data Center
Data Management