Edge computing continues evolving exponentially, and it’s never been more obvious after I served as a delegate for Gestalt IT’s Edge Field Day #2 (EFD2) in Santa Clara, CA. After returning home from Oracle Cloud World 2023 in Las Vegas just a few weeks back – more on my experiences here – I had the chance to catch up with Stephen Foskett’s team at GestaltIT and with many of my fellow delegates from past Tech Field Day events.
We had two solid days of presentations from three vendors who focus on providing all the resources that modern IT operations need for their edge computing needs. I found it particularly refreshing that each vendor took completely different approaches to solving the atypical challenges of modern edge computing.
Solidigm: Computing At the Edge Means More Data Locally
The team from Solidigm led off our event. I’d already heard about their offerings as a delegate at Storage Field Day #24, but it was interesting to see how they pivoted their technology pitch towards the Edge. One evident key trend is that as edge computing resources continue to grow more powerful, more analytics are happening closer to the edge.
Thus, more data is being collected, preprocessed, filtered, analyzed for relevance, and retained locally for eventual transport to centralized data centers. Solidigm already offers high-capacity SSD storage solutions, but they’ve also created purpose-built endpoint servers that can provide edge computing capabilities to handle this new paradigm.
We also heard quite a bit about how Solidigm has creatively evaluated the robustness and survivability of their SSDs through some great stories on testing them in extreme situations, including a railcar bed, attached to a bowling pin, and even – how appropriate, as the 2023 World Series draw near! – during a pickup baseball game.
StorMagic: Can I Get a Witness?
The StorMagic folks focused on different set of real-world edge computing challenges: how to keep all the nodes on the Edge communicating with each other in small-scale settings without breaking the bank to keep them synchronized. That’s often harder than it sounds and potentially quite expensive – especially for, say, a moderately-sized additive manufacturing facility with a smaller IT budget than its fully-capitalized counterpart.
That’s where StorMagic’s SvSAN solution shows potential: It needs only two servers in its cluster per location, and provides concurrency within the cluster through their Witness as Service component. During their live demonstration, that witness role was fulfilled by a Raspberry Pi – a considerably inexpensive and lightweight component that prevents split-brain syndrome within the cluster.
StorMagic’s offering thus appears priced attractively for SMBs – potentially within a $10K upper limit when factoring in reasonably resilient hardware. It also comes with a simple client UI for deploying and managing the cluster’s networking and hardware that’s useful for deploying applications and upgrades across the enterprise.
NodeWeaver: The Edge Ain’t Beanbag.
Showing off their bare-metal nanocloud concept designed specifically for modern edge computing, the folks from NodeWeaver focused on some stern realities of deploying, monitoring, and upgrading edge devices as well as the applications they run.
NodeWeaver definitely brought their A-game to EFD2, including a full-blown server and several different edge devices for us delegates to experiment with. The smallest of these relied on ATOM technology, so we’re not talking about super-powerful devices here. However, that wasn’t the point of what they showed us; instead, their technology is concentrated on handling some tough realities of edge computing, including the need to deploy networking and applications across a wide variety of diverse devices running a variety of operating systems.
I had the chance to observe an example of a typical NodeWeaver deployment to multiple devices, and even experiment with a tiny nano-node. Since I’ve seen my share of unexpected hardware and networking failures over the years, I decided to have some fun as well. To simulate what could happen purely by accident in a real-world office environment, I deployed my dreaded PCL test: I unplugged the node’s power cord halfway through the configuration.
As the NodeWeaver folks explained, their nano-cloud solution is designed to recover automatically in this scenario. Indeed, I did receive notification via email that although the initial configuration attempt had failed, it was resuming at an appropriate checkpoint; it eventually completed the configuration within a matter of minutes. That’s crucial, of course, for alleviating serious heartburn when deploying radically different devices, perhaps half a continent away.
(PCL, by the way, stands for Polish Cleaning Lady. They are standard issue in most Chicagoland office buildings. They are friendly, hard-working, and devoted to clean office spaces, so they aren’t afraid to unplug a device if it gets in the way of any cleanup task. This actually happened to one of my mentors many years ago; he still uses that story to explain how an Oracle database performs complete recovery after an unexpected instance termination.)
Ignite Sessions: Our Turn!
I also had a new Tech Field Day experience: Stephen Foskett offered several delegates a chance to show off our presenting chops. My colleagues and I talked about a plethora of topics: biases to beware in generative AI, the difficulties of security within edge computing, and even how to build your own chatbot leveraging retrieval augmented generation.
Those of you who’ve seen me present at conferences know how much I love entertaining my audience while trying to impart some new ways of looking at things, and I hope they enjoyed my bit of fun as much as I did showing my new it. (Spoiler alert: Beware the Forer Effect.)
EFD #2: Wrapping It Up …
I’ve yet to attend a Tech Field Day event that didn’t expand my brain, and EFD2 gave me plenty of new use cases, technology, and business risks to ponder – something I’ll definitely use to my advantage as I’m preparing new presentations for upcoming meetups and conferences. And I’m still hoping we can pull together a Data Field Day themed event sometime in 2024. With the incredible changes we’re seeing in our industry as vector databases and generative AI become mainstream topics, there have got to be plenty of new service providers who’d love to tell their story.
