Cloud Database News from Snowflake, SingleStore, Oracle, and Neon
As some vendors ride new momentum, one industry leader faces "fading" influence
It’s the beginning of summer here in New York, so I thought I’d share a few “hot takes” on recent developments in the world of databases. In the week ahead, I will be traveling to San Francisco, my third trip to the Bay area in the past couple of months. There’s so much happening in this fast-paced industry. Here’s the latest.
Snowflake takes on transactions
There were a handful of announcements at Snowflake Summit on June 14-15, including new capabilities for the company’s forthcoming Snowpark for Python environment (which is in public preview), access to data stored in on-premises storage (in private preview), and a Native Application Framework for building apps that are hosted in Snowflake’s Marketplace (private preview), which sounds a bit like Apple’s App Store model.
However, the biggest news, at least in terms of cloud databases, is Unistore, a capability that brings together transactional and analytical workloads in a single Snowflake platform. This had been the buzz in the industry leading up to Snowflake’s event in Las Vegas, and it’s the latest example of transaction + analytics convergence.
AWS, Google Cloud, Microsoft, Oracle, SingleStore, TileDB, and others have their own transaction + analytics databases. They employ different database architectures and are described in various ways—hybrid transactional analytics processing (HTAP); converged database; OLTP/OLAP; or simply universal database. But the idea is pretty much the same, which is to make it faster, easier, and more efficient to run queries and analytics on business data as it’s happening.
Snowflake has now latched onto this idea, and it’s a game changer. As part of the Unistore announcement, Snowflake introduced a new table type called Hybrid Tables (in private preview), which can be joined with existing Snowflake Tables for a holistic view across all data. According to Snowflake, customers will be able to use Unistore and Hybrid Tables to build transactional applications “with the same simplicity and performance” they expect with Snowflake and “analyze both transactional and analytical data together in near-real time.”
Further reading from Snowflake: “Introducing Unistore, Snowflake’s New Workload for Transactional and Analytical Data”
SingleStore introduces Workspaces and other new features
Speaking of universal databases, SingleStore this week announced a half-dozen new capabilities for its distributed SQL database, SingleStoreDB.
The new features include Workspaces, which are a way to isolate workloads within the database, providing compute scalability while minimizing data movement. Also new is a WASM (WebAssembly) Code Engine for porting external code into SingleStoreDB; Flexible Parallelism for faster query performance across CPU cores; expanded data access controls; a new Data API; and a dbt adapter. We will hear more about the new capabilities at the company’s upcoming (R)Evolution 2022 event, which I plan to attend.
If it sounds like a lot of product engineering is underway at SingleStore, that’s no coincidence. SingleStore recently hired two Microsoft veterans to give it more development firepower: Shireesh Thota joined as SVP of Engineering, and Yatharth Gupta as VP of Product Management.
SingleStore seems to have momentum. Earlier this year, SingleStore was named to the Cloud Database Report’s Cloud Database Top 20 ranking. For more, check out my recent interview with CEO Raj Verma.
Oracle repackages hybrid/distributed cloud services
Oracle has announced two Oracle Cloud Infrastructure solutions that are intended to make OCI deployments more feasible for a wider range of customers.
The existing OCI Dedicated Region is a complete Oracle cloud region for on-premises deployment. Oracle is now packaging OCI Dedicated Region with a smaller footprint (60-70% less space) at an “entry level” price of approximately $1 million annually. Oracle also previewed OCI Compute Cloud@Customer, a rack-scale solution that is aimed at even smaller environments than OCI Dedicated Region.
Yet, even as Oracle extends its cloud services to more customers, it’s under pressure in the database market. In April, I reported that Oracle had slipped from its #2 position in the database market to #3, behind #1 Microsoft and #2 AWS. That was according to Gartner’s market revenue recap for 2021. Now, Bloomberg has added color to this story of market disruption based on interviews with several businesses, including Shutterfly, JP Morgan Chase, and Nasdaq, that chose non-Oracle databases.
Bloomberg made note of Oracle’s declining market share, and observed: “Oracle’s influence is slowly fading.” That is probably true in the cloud database market, but you could also say that Oracle’s influence is growing in the healthcare industry with its acquisition of Cerner. And that may now that’s the bigger opportunity for Oracle.
Further reading: “Oracle’s Database Dominance Eroded by Cloud Rivals,” Bloomberg
Neon previews its PostgreSQL cloud database
Neon, a startup, has joined the fray of PostgreSQL database providers with its new serverless PostgreSQL database, called Neon. All of a sudden, the PostgreSQL market is getting a bit crowded. In May, Google Cloud introduced AlloyDB, its own PostgreSQL offering.
In a blog post, CEO Nikita Shamgunov explains that Neon is not yet GA, but users can get on a wait list. “We expect to open it up for everyone soon,” he writes. Apparently, HackerNews preempted the official launch of Neon, so Shamgunov is catching up to his own news, as sometimes happens.
“We think we have an opportunity to define the standard for cloud Postgres,” he says. “We carefully designed our storage focusing on cloud independence, performance, manageability, DevX, and cost. We chose the most permissive open source license: Apache 2.0 and invited the world to participate.”
Interesting footnote: Shamgunov was co-founder of MemSQL, which changed its name to SingleStore in 2020.
Further reading from Neon: “SELECT: ‘Hello, World’”
TechTarget interviewed KetteQ, an early adopter of Neon
Update: New data reveals what happens after a shooting
This is not about cloud databases, though modern data management should be part of the solution. At the beginning of June, I put out a call to action for our industry to rally around the need for more and better data to help policymakers understand and respond to the epidemic of gun violence that is causing so much loss and heartache.
Much of the effort to date has come from grassroots organizations and the media, the latest being a New York Times article that analyzes what happened in the immediate aftermath of 433 active shooting attacks. The Times reports that more than half of those situations (249 to be exact) ended before the police arrived. Which raises questions about the role of bystanders in active shooter attacks, and what works in terms of intervention. Using data from the Advanced Law Enforcement Rapid Response Training Center (ALERRT) at Texas State University, the Times’ analysis revealed that in 64 instances, a bystander subdued or shot the attacker.
This kind of data and analysis sheds light on the urgency of mass shootings, and we need more of it. If you’re in the data business and have ideas on ways to help, please contact me directly at jfoley09@gmail.com.
Further reading from the NY Times: “Who Stops a Bad Guy with a Gun?”