Dream Big, Go Bigger: Turbocharging PostgreSQL
September 12, 2023
Yugabyte CTO, Karthik Ranganathan just made a huge announcement!
OK, several huge announcements. He announced numerous product enhancements for YugabyteDB that are making all of our PostgreSQL dreams (and then some) a reality. The announcements ranged from groundbreaking bimodal query execution, letting YugabyteDB deliver PostgreSQL-like performance, to pain-free connection management, transforming a PostgreSQL weakness into a notable strength.
His presentation highlighted a wishlist of capabilities that will turbocharge PostgreSQL into the future via the innovative new features of YugabyteDB 2.19 and YugabyteDB Voyager 1.5.
Four Big Keynote Takeaways
Karthik challenged us to dream big, and then go ever BIGGER, when building a wishlist for PostgreSQL. Just as PostgreSQL delivered the dream application API, now it’s time to make the modern database of your dreams a reality.
Using the latest features of YugabyteDB 2.19, Voyager 1.5, and yugabyted, achieving your goals is now within reach. Here are four of the biggest points from Karthik’s keynote.
PostgreSQL Performance Parity Can Unblock Lift-and-Shift Modernization
The new capabilities provided in YugabyteDB 2.19 build upon the core strengths of PostgreSQL, expanding the types of applications that are an ideal fit for YugabyteDB. Thanks in part to the key innovations detailed below (and during the keynote), YugabyteDB now provides PostgreSQL-like performance for an even larger set of applications.
Distributed SQL has historically been a great fit for apps needing to scale (and scale big), but YugabyteDB now delivers both code compatibility AND performance parity to make it even easier to lift and shift large sets of applications from PostgreSQL to a modern distributed database.
So now organizations can enjoy the best of both worlds—the low-latency performance of PostgreSQL, combined with the built-in resilience and on-demand scale of distributed SQL.
Bimodal Query Execution:
YugabyteDB 2.19’s new distributed query engine (powered by co-located tables) helps enable the lift-and-shift migration of small and midsize applications to YugabyteDB.
Single-node PostgreSQL offers local execution for low latency, but no distributed execution. Classic distributed SQL offers distributed execution for resilience and scale, but with higher latency. YugabyteDB brings these two together for the first time. It introduces a bimodal query execution engine that supports both local execution for low latency and resilience. Then it easily switches (on a granular per-table basis) to a distributed execution engine for massively scale apps that require both resilience and scale.
Midsize applications running on single-node database instances can now effortlessly move to a fully distributed database environment. As the app grows, YugabyteDB shifts seamlessly to fully distributed mode to provide the massive scaling benefits of distributed SQL. Companies no longer need to change databases if and when their app needs to scale out.
Cost-Based Optimizer:
To further deliver PostgreSQL-like performance for any application, YugabyteDB 2.19 includes a new, distribution-aware Cost-Based Optimizer (CBO). The brand new CBO was built to fully capitalize on YugabyteDB’s unique distributed storage layer and includes support for query pushdowns, LSM indexes, and batched nested loop joins.
Underpinned by an innovative query optimizer testing framework, the new CBO now gets much closer to PostgreSQL by picking the ideal query plan for a wide range of benchmark tests. Testing shows that the new CBO can deliver up to 40% improvement in query execution time compared to previous versions of YugabyteDB. The CBO is available in preview today and can be enabled by setting two flags.
Pain-Free Connection Management Turbocharges PostgreSQL
During the keynote, we were reminded that (if we’re being honest) we’re pretty afraid of managing and maintaining connections in PostgreSQL. But this new update in YugabyteDB 2.19 promises to make us adore our connections again.
In a first for a distributed SQL solution, YugabyteDB 2.19 introduces built-in connection management, transforming a common PostgreSQL challenge into an advantage. New native connection pooling uses logical connections to guarantee consistent database connectivity for applications.When combined with our unique YugabyteDB Smart Drivers, YugabyteDB greatly simplifies application architecture, accelerating productivity.
This built-in feature makes it 20x faster to create connections and provides up to 5x more connections per node than standard PostgreSQL—all while maintaining the same throughput. Additional innovations in the new connection manager include pass-through authentication, native quality of service (QoS) controls, and simplified operations thanks to sticky connections.
Portfolio Assessment and Live Migration in YugabyteDB Voyager Reduce Migration Headaches
Data migrations—an often overlooked initiative we all wish were better—have been greatly simplified for many users with the release of YugabyteDB Voyager earlier this year.
Today, we are excited to release YugabyteDB Voyager 1.5, the latest version of our open source end-to-end migration tool, which now includes the following major enhancements:
- Live Migration for No App Disruption: YugabyteDB Voyager 1.5 now offers live, non-disruptive migration from Oracle with new optional fall-forward capabilities. Organizations can now easily move critical applications away from the complexity and cost of Oracle databases without impacting app availability.
- Built-in Portfolio Assessment: YugabyteDB Voyager 1.5 previews a powerful new migration assessment feature that scans available apps and databases. The detailed assessment provides insight into which applications, data, and schema are ready to be migrated—allowing organizations to accelerate their modernization efforts.
YugabyteDB 2.19 Includes Too Many New Features to Do it Justice in a Single Keynote
No matter how quickly Karthik speaks, there’s only so much he can cover in an hour! So here is a quick summary of some of the amazing updates that are all part of this new release. We’ll be sharing more details about these features (and helping you to unpack everything) over the coming weeks.
- PostgreSQL extension—pgvector—now in preview: YugabyteDB introduces initial preview support for pgvector, allowing vector similarity search and storing embeddings for applications needing these capabilities.
- Transaction semantics just like PostgreSQL: While YugabyteDB has always delivered industry-leading runtime compatibility with PostgreSQL, it now offers enhanced read committed isolation with pessimistic locking to round out support for key isolation features.
- Transactional Async Replication: High-throughput xCluster async replication minimizes RPO while maintaining referential integrity on replicas.
- Simplified Disaster Recovery Workflows: Promote replicas during a disaster or perform planned failovers periodically through an API or with a simple click of a button.
- Simplified OSS Experience: Use yugabyted to deploy and operate YugabyteDB clusters of any size, for global applications as well.
Concluding Thoughts
As Karthik concluded this packed keynote, we realized that the latest YugabyteDB updates are—no surprise here—checking off all those boxes on our “Dream Big…Go Bigger” PostgreSQL wishlist!
Find Out More
If you missed the DSS keynote presentation and want to know more about these exciting YugabyteDB developments, keep your eyes peeled for upcoming feature blogs and DSS replay videos.
Distributed SQL Summit
Thousands of people have registered to attend today’s DSS virtual conference with its world-class keynotes, insightful case studies, and technical presentations from both YugabyteDB database and industry experts.
If you have not signed up, there is still time! The Distributed SQL Summit runs from 9 am until 2:15 pm PDT today, and you can register here for free. A quick and simple registration will ensure that you have unlimited access to all the replay videos, demos, and presentations.
