Tech Talk

Data at the Edge and the Role of Memory and Storage

Edge computing


Last updated 26 April 2022

According to IDC, $274bn will be spent on edge computing in 2025. This, in comparison to IDC saying $175bn would be spent on hardware, software, and services for edge solutions. Quite the increase.

So, it makes sense we start discussing the impact edge solutions will have on memory and storage as part of the ongoing digital transformation which continues on a steep upwards trajectory of change.

 

What is the edge?

 

Edge computing can take up many forms. It places computing power, storage, and analytics closer to where data is produced, utilising sensors and IoT devices to collect data, process it, and feed it to edge services, or to the cloud. 

So why is this happening? Firstly, centralised cloud computing will be a thing now and long-term. It’s part of the solution. But in placing such computing and storage as close to the devices creating data makes it achievable to gather true real-time data there and then, answering our questions with immediate effect – be that super low-latent transactional decisions or safety choices within autonomous vehicles which lives can depend on.

 

What do we mean by data at the edge?

 

We’ve looked at the spend by 2024, so let’s look at data. What we will see generated by 2025 in terms of data will be 175 zettabytes, which is 10x the amount of data we were seeing in 2016 (as per IDC in 2018). 

Such a volume of data will need to be distributed to which edge data centres come into play, with Gartner saying that in 2022 around 75% of all data will need work and action at the edge. Managing data at the edge, such data centres are located close to the edge of a network. The edge data centre sits in between the IoT devices of today and the public cloud or your centralised data centre.

Ultimately, data at the edge comprises infrastructure, services, and software between cloud/data centres and ‘intelligent’ endpoints. 

 

 

A way of looking at the edge as per IDC

 

The benefits and challenges of data at the edge

 

It's easy to see the benefits of edge from the above, but let's look at some unique benefits:-

  • Real-time - as we mentioned, edge is all about real-time and low latency. Large amounts of data can take time to transfer over further distanced, no matter how fast our 5G is, or our general network connection. With edge, the latency is reduced for edge applications depending on information in real-time.
  • Costs - with anything that takes longer, it will likely cost more. The same can be said with bandwidth for large volumes of data moving over longer distances. With computing at the edge, costs are reduced.
  • Location - The proximity of edge technology keeps sensitive data and information closer to the source. This can be important for compliance and data law, something key for most businesses these days.

A challenge comes in the shape of managing vast amounts of distributed data. Supporting data at the edge which is deployed in geographically diverse points of presence (PoPs), needs planning and monitoring at a high level.

What’s more, in terms of security edge computing and holding data at the edge can be seen as risky, or a threat. Traditionally, data is held securely in a centralised cloud-based data centre system but spreading data at the edge and across various devices needs a greater understanding of potential security issues and ensuring maximum protection.

 

How memory and storage can meet edge demands

 

Speed

 

Data at the edge requires millisecond responses. With the device itself processing data at the edge without the need to send data to the cloud for remote processing, device manufacturers design in CPUs and FPGAs to do faster neutral processing.

With this, memory can meet edge demands through high-speed memory and storage can do the same with high-capacity SSDs to store and serve data in edge devices.

 

Security

 

Outside of speed and capacity demands, peace of mind to adopt edge for security is a demand within itself. This is something SSDs can meet. And what’s more, edge servers only choose solid-state-drives. 

SSDs provide a guarantee in terms of reliability when it comes to losing or hindering data in form of impending cyber-attacks. What’s more, when it comes to edge, ECC technology for error correction at high-speed transmission to ensure data accuracy and access is key when it comes SSD in edge. Something the latest technologies can provide.

What's more, SSDs come with other security features to protect mission-critical data at the edge including data sanitisation options like crypto erase, fast erase, normal erase and fully customisable data sanitisation; something that is definitely of need for the industrial industry. 

The same can be said about ECC on RAM (DRAM), helping play a critical role in data for its automatic ability to detect and correct memory errors to fight data corruption at the edge.

 

Environmental 

 

Whether it be a data centre or server, or industrial, putting data at the edge can mean working with tough, challenging conditions and remote settings. With this, edge devices and edge SSDs must be engineered with a rugged profile. Edge server SSDs have to be able to cope with extreme environments and challenges to deliver enterprise-class QoS and ultra-low latencies. 

SSDs are able to do this when designed with features like conformal coating, which protects the SSD from moisture and contaminants, delivering that rugged outcome. What's more, thermal control features can tackle issues that include overheating and thermal throttling that is pretty much unavoidable in high-performance PCIe NVMe flash storage devices and is totally common in edge servers. 

Further to this with devices at the edge, they can suffer from unexpected power loss of system failure. So how do you monitor the health of such devices?

Therein lies a couple of solutions from Taiwanese manufacturer Innodisk who offer an SSD called InnoOSR or On Site Recovery. InnoOSR’s unique architecture allows operators to restore the InnoOSR SSD drive to a working state with a single click of a (physical) button – or through other innovative ways depending on how the operator integrates InnoOSR in their application. 

Alternatively Innodisk offer something called InnoAGE™  Industrial Grade SSD, which is designed for both in-band and out-of-band network management, providing full recovery even when the operating system has crashed or is severely impaired to the extent that in-band management would be of little help. System operators can quickly revert to the default settings from the cloud-based dashboard in the case of a device or system crash. 

When looking to select and source RAM or storage for your edge set-up, speak to Simms. We can advise on the best memory and storage for such environments from our vendors in this realm which include Micron, Kingston, Innodisk, ATP and Solidigm.

Enjoy the video below which explains through a blackboard-tutorial how critical SSDs for success in edge data centres and computing: 

 

More resources

 

As mentioned above, we are working with world-leading vendors and innovators in memory and storage. To help inform and highlight features further, we have added a selection of worthy resources below to draw from.

 

 

 

Author

Drew

Drew is Marketing Lead at Simms, leading our marketing department. Drew has a strong knowledge datacentre and server proposition, and leads on our industrial and embedded side of the business also.