ACID Compliance for the PDE Exam: Strong vs Eventual Consistency

ACID compliance is one of those topics that sounds dry on paper but shows up on the Professional Data Engineer exam in ways that actually matter for picking the right GCP database. If you can read a scenario and decide whether the workload needs full ACID guarantees or whether eventual consistency is fine, you have already eliminated half the wrong answers on a chunk of the database questions.

Let me walk through what ACID actually means, how strong and eventual consistency differ, and which GCP services land on which side of the line.

What ACID stands for

ACID is a set of four properties that guarantee reliable transaction processing in a database. Every Professional Data Engineer candidate needs to be able to recite these and recognize them in scenario language.

• Atomicity means all or nothing. Every transaction either completes entirely or not at all. If one step fails, the whole transaction rolls back and the database is left unchanged. No partial writes, no half-applied updates.
• Consistency means data integrity is preserved. Every transaction takes the database from one valid state to another, respecting constraints, triggers, and cascades. The data written follows all the rules defined on the schema.
• Isolation means transactions do not interfere with each other. Even when many transactions run at the same time, each one behaves as if it were the only one running. The end result is the same as if they had executed one after the other.
• Durability means data is permanent. Once a transaction is committed, it stays committed, even if the server crashes or loses power right after.

That is the whole acronym. Atomicity, Consistency, Isolation, Durability. Reliable, all-or-nothing, ordered, permanent.

Strong vs eventual consistency

Consistency in the ACID sense is one of those four properties, but consistency also shows up as a separate axis when you compare databases. Specifically, you will see strong consistency and eventual consistency as two different models for how a distributed system handles reads after writes.

Data consistency is evaluated on two criteria. Timeliness asks whether the data being read is up to date. Order asks whether the sequence of updates is preserved correctly.

With strong consistency, every query is guaranteed to return the most recent update. All parallel processes have to apply changes in the same order, which means the system synchronizes carefully before responding. The trade-off is speed. Strong consistency can slow down query response times because the system has to coordinate. You want this model for anything where accuracy is non-negotiable, like financial transactions, payment processing, or inventory counts that drive a checkout flow.

With eventual consistency, the system prioritizes faster responses and lets some updates land out of order or with a short delay. Over time, all updates propagate and the data converges to a consistent state, but in the short term a read might return stale data. This works well when speed matters more than immediate accuracy. A nationwide census or a large-scale analytical workload tolerates eventual consistency cheaply.

The mental shortcut for the exam is that ACID compliance entails strong consistency. If a scenario calls out ACID, it is also calling out strong consistency by implication.

Why ACID matters on the exam

ACID compliance shows up on the Professional Data Engineer exam most often as a requirement embedded in an industry scenario. The industries that drive these questions are predictable.

• Finance needs ACID for transactional data like bank transfers, payments, and trade orders. Not every piece of financial data needs full ACID, but anything that moves money does.
• Healthcare needs ACID for patient records, prescriptions, diagnoses, and billing. The integrity of the medical record is the whole product.
• eCommerce needs ACID for order processing and payment confirmation. If a customer pays, the order has to update, and both have to succeed or fail together.

Government data like legal records and tax filings also lives in this bucket. When you see any of these contexts paired with words like transaction, payment, order, record, or audit trail, ACID is on the table.

ACID compliance in GCP services

This is the part of the topic that turns into direct exam points. You need to know which Google Cloud database services are ACID compliant and which are not.

• Cloud Spanner is fully ACID compliant. It is a globally distributed relational database with strong consistency at scale. If the scenario says global, relational, high availability, and ACID, the answer is Spanner.
• Cloud SQL is fully ACID compliant across all three engines it supports, MySQL, PostgreSQL, and SQL Server. This is the default pick for traditional transactional workloads that fit on a single regional instance.
• Firestore is ACID compliant at the document level. It is a NoSQL document database, so the ACID guarantees apply within a document and within transactions you explicitly open, which is great for app backends that need transactional integrity but want the flexibility of NoSQL.
• Cloud Bigtable is not fully ACID compliant. It is built for very large scale, low latency, and wide-column workloads, often analytical or time-series. If a scenario rules out transactions and asks for scale and speed, Bigtable fits.
• BigQuery is not on the ACID list either. It is an analytics warehouse, not a transactional store. If a question pairs ACID with BigQuery, that is almost always a trap answer.

The pattern to memorize is that Spanner, Cloud SQL, and Firestore handle transactional workloads with ACID guarantees, while Bigtable and BigQuery are for analytics and scale without full ACID.

If you can match the workload to the right consistency model and then to the right service, you have built the muscle that the Professional Data Engineer exam is really testing on this whole family of questions.

My Professional Data Engineer course covers ACID compliance, consistency models, and the full GCP database selection framework you need for exam scenarios.