VoltDB was established based on the work of Michael Stonebraker, distinguished data scientist, whose work has spawned a number of database companies, starting with Ingres (now Actian). Stonebraker was the 2014 winner of the “Turing Award” from the ACM (Association for Computing Machinery), and which stated, “He is the inventor of many concepts that were crucial to making databases a reality and that are used in almost all modern database systems”.
We recently had the opportunity to speak with Peter Vescuso, CMO of VoltDB, and also to sit in on a webinar about the IoT by the company’s CTO, Ryan Betts.
Distinctive Technology Approach
It is clear that VoltDB believes that its technology is uniquely suited to the emerging world of “fast data” – as distinguished from “big data” – and that the proliferation of fast data is being driven at an ever-accelerating rate by the growth of the IoT. According to Vescuso, they have unique technology.
The bulk of their business has been in the telecom, financial services and media/entertainment, as well as the growing IoT segment. In telecom/mobile, their largest vertical, they deal primarily with providers to the carriers, such as Openet and HP, as well as Nokia and Ericsson, who offer software support systems such as OSS/BSS (operations support systems, business support systems.)
Openet, a supplier of software support systems to telcos, for example, found that as data usage in its clients’ networks exploded and the telcos moved to virtualization and the cloud, existing database solutions were inadequate. It called VoltDB’s approach, “remarkable and absolutely groundbreaking.” It cited advantages in combining data resiliency, redundancy, speed, and processing, which allowed telcos to marry real-time data streams with historical data for rapid new insights.
Designed “Almost Purposefully” for The IoT
Vescuso describes the company’s technology as “almost purposefully designed for the Internet of Things.” He describes it as, at a high level, being super fast, designed for virtualized cloud environments and preserving data integrity. He states that “we need next generation apps” to demonstrate the superiority of their approach. This will revolve around sensor data management of the IoT. VoltDB’s system can ingest, process and allow query of the data in realtime. He states that other DBMSs may sometimes actually lose data.
He recounted that VoltDB is already starting to receive significant IoT-related contracts, where fast data is a critical consideration. For example, in late 2015 the company announced a deal with Mitsubishi Electric to support data ingestion and analysis for six million smart meters (or AMI – Advanced Metering Infrastructure – meters) in Hokkaido and Shikoku provinces in Japan.
He cited another deal where VoltDB was selected by CGI to power a data management project for over 50 million meters, including electric, water and gas, in the UK. The expected benefits include cost control, managing authorization issues, monitoring compliance and improving energy efficiency
Another major project involves a client who he describes as a major data storage company (not yet publicly identified) that monitors sensors worldwide for temperature and other readings and is using VoltDB’s DBMS along with predictive analytics.
Future of IoT Architecture
In a webinar related to a “Unified IoT Architecture,” VoltDB CTO Ryan Betts delved into the issue of the balance between processing sensor data, “at the edge” versus “in the cloud.” Betts devoted considerable discussion to “fog computing,” a term popularized by Cisco and IBM to describe one version of edge computing.
The basic concept is that with the torrent of data coming from distributed devices of the IoT, there will be situations where it is inefficient to try to transmit all of the data to the cloud and it will be necessary to locate a good deal of processing capability at, or closer to, the edge where the originating sensors reside.
Especially where latency or cost of transmission are factors, Betts emphasized that edge processing might be the only viable alternative. He mentioned that a good deal of data from sensors, which cannot be efficiently handled in the cloud, is actually lost. Betts addressed these issues by referring to the more formal division of IoT computation requirements into three categories which is sponsored by the Open Fog Consortium:
1. “Regulatory” Control – which involves the monitoring of devices.
2. “Supervisory” Control – involving setting the goal state, e.g., temperature for a thermostat.
3. Decision Support – provides the feedback loop and models.
In general, he stated, the regulatory and supervisory functions are localized and could be done at the edge, with the decision support in the cloud. However, he also noted some of the reservations about edge or fog computing, such as the fact that it introduces an additional tier of compute, storage and networking elements, as well as power management issues, and that it is complex and could be expensive.
VoltDB has been chided for its approach to raising financing from the venture community, an approach Vescuso describes as “deliberate and conservative.” This used to be regarded as a sign of strength and prudence for tech companies, although in today’s world of artificially inflated financial markets, the race to maximize fund-raising has become a badge of success for many companies, some of whose business models we seriously doubt.
Vescuso also points out that the company’s financial soundness is bolstered by long term customer agreements that provide financing; as well as by money received from certain strategic investors who are paying the company for development help on some long term projects.
While not attempting to make technical judgments on the company’s DBMS products, we are favorably impressed by their progress in the marketplace. We believe that VoltDB is a company to be watched particularly because of the optimization of their technology around the data streaming and analytics requirements of the IoT, which are often going to necessitate almost instantaneous responses.
We don’t have a strong opinion about the lasting importance of Fog Computing as a specific approach to the issues of balancing the need for processing at the edge, versus the cloud. However, we are extremely interested in issues that relate to the evolution of sensors and smarter devices at the edge.
The current view of the IoT has been driven largely by cheap, basically dumb, sensors and includes the following four key elements or assumptions:
1. Low-cost edge elements: As Goldman Sachs listed as the first factor driving the IoT – “Cheap Sensors” (“The Internet of Things: Making sense of the next mega-trend,” GS 9/14).
2. Higher broadband capabilities will ubiquitously evolve quickly.
3. Big data farms can handle it all.
4. Intelligent analytics will devolve from the big data farms.
With more intelligent sensors, the edge will connect to the rest of its IoT family by communications, which reflect a stream of succinct information that has been algorithmically processed and contains both contemporary and predictive intelligence.
This approach could greatly reduce reliance on several key assumptions relating to broadband requirements and Big Data analysis. This includes the underlying latency thresholds and even the functionality/survivability capabilities.
Visit their website: www.voltdb.com