Spanner

All reads see the latest write, period.

Systems like Google Spanner or a single-node PostgreSQL database provide this.

This is the ultimate CP choice, but the performance cost is massive because every read and write requires complex distributed consensus.

consensus.So engineers try to slide down the spectrum to gain performance and availability.

First, in AI and Agentic Systems, we explore model context protocol trace spans in CloudTrace, direct publishing of BigQuery data agents into Gemini Enterprise, Gemini 3.8 Flash arriving by default, and automated client-side log streaming in Gemini Enterprise Agent Platform Workbench.

Second, in data and analytics, Spanner unveils both Spanner queues in general availability and automatic query literal parameterization.

We will also dive into Cloud SQL introducing the C4 machine family alongside asynchronous point-in-time recovery during disaster recovery failovers, Bigtable protobuf schema bundles, BigQuery inline aggregate filters, and major updates across Looker and Looker Studio.

Third, in containers and compute, Google Kubernetes Engine boosts pod density up to 512 pods per node and introduces an early preview of GKE ambient networking for sidecarless service meshes.

Finally, the legacy video intelligence API has been officially deprecated.

It will be shut down on September 14th, 2027, so teams should begin planning migrations toward Gemini multimodal vision capabilities.

Let us pivot from AI models to the data tier, where we have massive news across BigQuery, Spanner, Cloud SQL, and AlloyDB.

Starting with BigQuery, SQL developers get a brilliant syntax feature in preview.

But in fact, we're the opposite.

We give you the union of all clouds because now you can write the code with a native SDK for Google, for Cloud Spanner, for Amazon, for SQS, for Azure, for cloud, for let's say, Antra ID or Cosmos DB or whatever machine learning service that is on Azure.

And we allow you to consume those services without credentials.

Why is it important in the era of AI? You do not want to leak to the LLM the credentials.

I also lead operational databases for all of Alphabet.

And so whether it's Gemini or Gmail or YouTube, all of these run that same database infrastructure on the same set of systems in Spanner and Bigtable.

In many ways, my second job is very much like the interlocutors of my first job, people who build and run infrastructure teams.

They run it on behalf of various businesses.

There's been, you know, traditional databases where scale-up systems.

And so while they evolved coming to the cloud in a very evolutionary path, there were also these, what I call revolutionary systems, scale-out systems.

Bigtable was one of them that we built in Google about a little over twenty years back, and Spanner is another one of them.

And these set the tone for a whole industry where people approach this data management problem from a fundamentally different way.

Now, I ran into an interesting post recently that noted modern data platform can have a bunch of actually different systems underlying it.

An example might be that your software or your data platform could use Postgres, Mongo, Cassandra, ClickHouse, and maybe even some new SQL like Spanner or Cockroach.

Now, some of you might think that's not reality.

It's not gonna happen in my organization.