HarperDB’s Journey on Verizon 5G Edge: Collapsing the stack in a 5G world

by Jaxon Repp & Margo McCabe

As an enterprise database, HarperDB’s engineering and product teams eat, sleep, and breathe precisely two things: simplicity and performance. We optimize queries. We optimize storage algorithms. We optimize intra-node communication. We wake up in the middle of the night thinking about whether quantum entanglement would be a viable data replication strategy (best current guess: no).

At a certain point, though, the product is no longer the bottleneck — it’s the network. It’s the distance between the client and the application or between the application and the data that drives it. By reducing the distance data has to travel, we know we can save several orders of magnitude more time for a transaction than we can by optimizing our storage algorithm.

That’s why the expansion of 5G Edge is a big deal for HarperDB — and application performance in general. Mobile network infrastructure moves compute resources closer to clients on the network, reducing the latency associated with hops across the public internet. The performance improvement is dramatic. We’ve seen queries that took 5 seconds on a monolithic database in a region thousands of miles away take only 1 second when run on HarperDB Cloud’s distributed platform. But those same queries might take just 50 milliseconds on a 5G Wavelength instance due entirely to reduced proximity and network complexity.

Using Verizon 5G Edge with AWS Wavelength, HarperDB offers lower latency, greater stability, and reduced complexity for data-intensive projects as they move to the edge. And because we integrated it into HarperDB Studio, our customers can easily scale their architecture by adding HarperDB nodes closer to their customers with a few clicks. We’ve delivered game-changing performance for customers in the gaming, industrial IOT, transportation, and retail sectors.

Why The Edge Matters
Shifting computing and application workloads directly to the edge enables better and faster decision-making. It also eliminates the need to transport massive amounts of data from where it is accrued (out in the real world) to the cloud for eventual analysis. Most sensor data is useless to the organization and consumes large amounts of network and cloud resources.

Cloud-first data architectures end up being more expensive and less performant than edge-first solutions for several reasons:

• Vertically scaling infrastructure for performance can become incredibly expensive
• Performance will always be inferior when data is farther away
• Ingress/egress costs for traffic between endpoints and database
• Allowing applications to work and persist data locally is faster and more stable

It seems like a no-brainer that the edge is a new way to optimize, and 5G Edge with Wavelength comes with the added benefit of reduced network complexity. But while the benefits are clear in theory, some organizations are hesitant to make the shift. The challenge of migrating applications and data closer to customers seems daunting. Still, companies like HarperDB and infrastructure providers like AWS and Verizon are delivering turn-key solutions that meet the needs of today’s workloads.

Verizon Wavelength and HarperDB Cloud
To help enable customers to make the leap, HarperDB has integrated AWS Wavelength offering into HarperDB Cloud. The integration allows developers to leverage HarperDB’s application development platform and SQL/NoSQL persistence engine on high-performance edge servers integrated with Verizon’s network. Applications running on AWS Wavelength benefit from significantly reduced latency, dramatically improving user experience and increasing reliability. HarperDB at the edge simplifies development, collapses the stack, and reduces complexity to lower TCO when compared to traditional application architectures.

HarperDB is the first third-party turn-key as-a-service solution running on AWS Wavelength platform. As a result, customers can reduce latency (by shifting APIs to the edge), lower costs (by eliminating the need for separate application servers), and increase performance (by integrating core HarperDB access and eliminating the hop from the API to the database), with all the convenience of a managed service running on the largest and most stable infrastructure platform in the world.

Customer Results
The HarperDB & AWS Wavelength solution recently made our customer’s smart edge application 250 times faster and more cost-effective. Minimal application refactoring is required, and API results are globally cached in HarperDB on AWS Wavelength.

HarperDB’s Custom Functions development platform was used to create an API caching layer that can reduce latency from upwards of 5 seconds down to 20 milliseconds or less. The returned value is clustered across all AWS Wavelength servers, and servers will return the cached value from HarperDB. Since HarperDB’s data storage replicates cache globally in less than 110ms, this solution also reduced the number of calls to the source API by 99%, which in some cases also reduced their costs with those API providers.

Provisioning Wavelength instances through the HarperDB Studio dramatically minimizes the number of external API calls. Shaping the data to reduce processing down creates a fantastic user experience coupled with rapid speed and real-time UI updates.

Getting Started with HarperDB on Verizon Wavelength
Follow these instructions to spin up a HarperDB instance on AWS Wavelength using HarperDB Studio.

To spin one up using your existing DevOps process, use this cloud formation template.

Looking Ahead
What an exciting time! Developers now have more choices than ever regarding where data can live and how it moves around the environment over time. Choosing a solution closer to your customers offers unparalleled performance and flexibility, and lowering costs seems like a slam dunk.

At HarperDB, we prioritized integrating our 5G Edge with Wavelength offering because optimizations outside our core product are just as crucial as optimizations within it. The improvements offered by this solution have delivered incredible results so far, and we’re eager to see where our customers will take it next.

We strive to be the obvious choice for customers with massively distributed user bases, including network providers, social networks, and gaming companies. We’re always looking for new ways to improve performance and flexibility, and we’re excited to be launching user-defined modules soon. This “features-as-code” architecture will allow any application to be deployed, managed, and empowered by HarperDB’s distributed persistence engine. And since speed and flexibility don’t carry much weight if you can’t get it to work on your infrastructure, we also spend lots of time developing DevOps templates, integrating partner technologies (like this one), and exploring enterprise tooling like AWS Outposts.

Availability
HarperDB’s AWS Wavelength offering is available now. Users interested in delivering ultra-low-latency performance through HarperDB Cloud can spin up instances in HarperDB Studio (https://studio.harperdb.io). Pricing varies by instance type and location, with additional configurations and regions being added as they come online.