While the world was watching the coming together of the best soccer players in the world in the Rub'al-Khali desert, the biggest players in the technology world convened in the Mojave desert.
AWS re:Invent has become the World Cup of technology conferences. Being on the largest stages, temperatures were running high at both events, and not just because of the Europeans’ aversion to air conditioning.
SingleStore was at AWS re: Invent in full force along with our partners including IBM, RedHat, HPE, Zesty, Grafana and Confluent — as well as SingleStore customers who exhibited in the showcase such as Palo Alto Networks, Imperva, Goldman Sachs and CapitalOne. It was thrilling to see SingleStoreDB power these real-time applications and customer-facing analytics on the exhibit floor. We’d like to share some of the excitement, activities and our take on the announcements that happened, and some that didn’t.
First of all, in-person technology events are clearly back. With reports of close to 50,000 in attendance, it’s clear we are no longer living in the COVID-induced, all-digital world. The booth conversations, informal chats between sessions and serendipity of chance encounters with former co-workers, customers and partners is something that pure virtual events simply can’t match. But with hundreds of thousands attending the keynotes online, the event world’s future is forever hybrid.
SingleStore has been called “absolutely magical” and “the database of our dreams” by its users, so it was only fitting that we book our favorite magician for our re:Invent booth who managed to educate and awe the audience with SingleStore database “aha!” moments. Our solutions consultants help pull back the curtain and show conference attendees what SingleStoreDB can do for real-time applications with analytics using a real-time marketing analytics application which you can see here — and find the source code to run it yourself in a free trial.
Over at the Intel booth, another kind of wizardry was on display. Intel, Chooch.ai and SingleStore teamed up to demonstrate the power of real-time analysis of images and videos. SingleStoreDB has first-class support for relational, document, geospatial, time-series, full-text search and vector data.
Vector functions are particularly helpful in determining image similarity and performing matches on image and video data. In the booth demonstration, floating-point vectors were generated by Chooch.ai using a neural network trained to categorize the images. Then these vectors were streamed into a SingleStoreDB table where regular SQL queries executed vector functions of DOT_PRODUCT() or EUCLIDEAN_DISTANCE() to find the closest image matches. The execution of these vector functions are accelerated by SingleStoreDB’s support for Intel AVX2 SIMD instructions through SingleStoreDB compilation of every SQL query to machine code.
No AWS re:Invent happens without theCube — and for the 10th year running, they’ve provided in-depth coverage from the showfloor. We were fortunate enough to have two appearances on theCube during the event this year.
In the first, SingleStore’s SVP of Engineering Shireesh Thota joined Hemanth Manda, our IBM partner andHead of Strategic Partnership. The two spoke with co-hosts John Furrier and Savannah Peterson about the partnership between IBM and SingleStore, describing how SingleStore’s mission to unify and simplify modern real-time data has been a perfect match for IBM’s goal to simplify and broaden the use of AI.
Shireesh shared the recent steps to make this a reality, including the SingleStore support of IBM’s embedded AI through integrating the IBM Watson Natural Language Processing library and the custom connector to IBM Cognos. Hemanth emphasized IBM’s philosophy of building an open ecosystem and highlighted the partnership with SingleStore as a perfect example of this initiative in action. It seems no conversation these days about data can avoid the so-called “modern data stack”.
When asked by John to share his point of view on this, Shireesh observed that the modern data stack has too many layers and components. It’s too complicated and all those moving parts require an unnecessary amount of data movement in the form of replication and/or ETL. By combining multiple processing , storage and data models into a single unified system, SingleStoreDB accelerates the data experience for real-time applications and analytics while reducing cost through greater efficiency and simplifies the data infrastructure stack.
The interview closed with Shireesh providing a quick overview of the highlights of the SingleStore Winter Release — like an up to 400x increase in SELECTs on JSON data, and the full availability of embedded WebAssembly (Wasm) in all SingleStore product editions.
In the second SingleStore appearance, I sat down with Lisa Martin to discuss the AWS announcement of zero ETL, the role of the data analyst, real-time operational analytics and some of SingleStore’s most memorable billboards. Zero ETL is a new feature AWS announced to replicate data from AWS Aurora to AWS Redshift. Building and maintaining Extract, Transform, Load (ETL) jobs which combine data from multiple sources into a data warehouse can be a big headache. In the words of AWS CEO Adam Selipsky, zero ETL “helps solve one of the greatest ETL pain points for our customers”. This is a good move for AWS as it’s solving one of the growing data integration challenges of dealing with so many separate, specialized services.
From a SingleStore point of view, it is seen as another inefficiency from the previous era of cloud data technology and architectures. There’s been a 50-year divide of OLTP and OLAP technologies which SingleStore has unified to deliver the lowest latency, real-time processing for applications and analytics. A single, unified database technology that reduces data movement, reduces data duplication and simplifies application design and does this with a proven better total cost of ownership.
Besides integration between AWS services, another major theme was the announcements of new serverless capabilities. AWS Aurora Serverless v2 was announced earlier in the year, and AWS Neptune Serverless was announced a month before the event. What was somewhat of a surprise at this year’s AWS re:Invent was what was not announced. After launching new specialized NoSQL databases in recent years, no new NoSQL databases were launched this year.
In just the last four years, the AWS Database team has released AWS DocumentDB (MongoDB), AWS Keyspaces (Apache Cassandra), AWS Timestream (time-series database) and AWS MemoryDB for Redis. We’ve written quite a lot about the complexity introduced by so many specialized database technologies and the need for simplicity by eliminating data infrastructure sprawl. And, there’s been a rising chorus by industry analysts, including Stephen O’Grady at RedMonk, noting that the pendulum is swinging back to general-purpose relational SQL databases — away from the NoSQL database databases of the last 15 years. Andy Pavlo cited just last week that a relational SQL database is the best starting point for new applications, and in 2022 we’ve witnessed the shift of mainstream NoSQL databases like MongoDB trying to bolt on some sort of quasi-SQL capability.
Of course, to build a rock-solid, efficient distributed relational database engine with a good query optimizer is much harder and a lot more work than supporting a few selected SQL operators, and SingleStore has spent 10 years refining its system. The future is already here; it’s just not evenly distributed, as William Gibson wrote. That’s what we’re seeing in the database market today.
SingleStore and a small cohort of other modern distributed relational database companies provide much simpler and better tech to serve the needs of modern applications, yet we still see companies using the previous, Second Era approach. If you haven’t had a chance to walk through Cloud Data Evolution, watch it here.
What distinguishes SingleStore is our vision and our focus. Our vision is to unify and simplify modern real-time data. Simplicity is the ultimate sophistication, and the simplicity of SingleStoreDB in bringing together real-time diverse workloads across the spectrum of streaming, transactions and analytical without forcing hard choices or compromises is what makes SingleStoreDB sing — and is why we have such a vibrant and vocal community of developers and customers who talk publicly about it.
And with that, thanks for reading. We’d like to invite you to our next magic show. Join us to hear LiveRamp talk about their success with SingleStore as part of our Winter Release on Dec. 15.