Anyone who has read William Gibson, Neil Stephenson, or Neal Asher has known for a long time that edge computing was coming. People have been playing around with concepts like Beowulf clusters since the 90s. That said, technology is finally at a point where advances and uses of edge computing are beginning to overtake massive server environments.
At HarperDB, we are pretty excited about edge computing for several reasons. First, we are pretty big believers in the free flow of information. We believe that edge computing already is leading to, and will continue to lead to, a more democratized control of information. We see a day when peer-to-peer Internet connectivity is the norm not the exception.
A few years ago when I was at AWS re:Invent, I noticed a major trend of folks complaining about their massive AWS bills. At the time with my previous employer, we were spending roughly $50,000 to $70,000 a month with AWS so I felt their pain. As the cloud is becoming more and more expensive, people are looking at hybrid cloud models to reduce their cost on static workloads.
I think for some this model has merit; however, in other cases I see a different path. Today I am walking around with a Samsung Galaxy Note 8 in my pocket, which has more computing power than an AWS T1 Micro. Companies have spent trillions of dollars on hardware like wearables, cell phones, IoT devices, weather sensors, satellites, point of sale systems, and other edge computing devices yet they are continuing to move their data workloads back into the cloud.
The combined computing power of the edge devices is well beyond what is being paid for at a king’s ransom in the cloud, yet people are vastly underutilizing these devices. These micro-computers are already distributed across geos and power sources and make themselves ideal for tackling redundancy and latency, yet people are moving their data back to a hand full of locations hosted by one of the major cloud providers.
Don’t get me wrong, I think that Google, Microsoft, and Amazon provide some incredible services on their clouds. It makes sense for companies to take advantage of services like Lambda, S3, Google’s machine learning services and much more. These are awesome products. That said, its crazy to pay cloud companies for static compute and storage over and over again each month, when companies have many times that much compute and storage at their fingertips that is being underutilized.
The reason is that while applications and middleware technologies over the last 10 years have done a fantastic job of becoming distributed, databases have not. If you want to run a database on a micro-computer, at this time you basically have a few options: SQL Lite, Couchbase, Postgres, or MySQL.
These are all great products and have awesome use cases. That said, if you want to run an enterprise scale cluster with billions of transactions and use your edge to process all steps within the data value chain, they are probably not ideal. Postgres and MySQL were not designed to run in an edge environment and from a resource perspective while capable, are not ideal. SQL Lite is a great product for embedded devices and is amazing for being used on the edge; however, it was not designed to replace your data warehouse or your core application database. As a result, the recommended pattern is still to move your data back to some version of the cloud.
HarperDB is 65mb. It can be run directly on micro-computing devices using the ARM6 or ARM7 build. It can be run in a server environment using the Linux build. It scales to your hardware automatically. It is designed to be an HTAP database (Hybrid Transactional/Analytical Processing) so it can handle large scale ingestion as well as large scale analytical processing in the same environment.
The enterprise edition of HarperDB contains clustering capability that with minimal configuration, facilitates the intercommunication of devices. This makes it simple to leverage existing hardware investments to do clustering on the edge. With HarperDB, companies can take advantage of the billions of dollars they have spent on edge computing devices and rather than simply use them for collection, begin to use them for processing and compute directly on the edge.
We think with this model rather than duplicate spending on capital expenditures, companies will begin to reinvest those dollars in innovation. We believe that by moving computing to the edge, companies can see massive cost savings while also democratizing the global flow of information. Try it for yourself today.