Batch Apex vs Queueable Apex: Which is better for processing 200k records under governor limits?

Answer to the Core Question

Batch Apex is the preferred approach for processing 200,000 records while staying within governor limits. It automatically splits the workload into execute chunks, each running in a fresh transaction, so SOQL, DML, and CPU limits reset per chunk. Queueable Apex, by contrast, runs as a single asynchronous transaction per job; processing the entire set in one job would hit the same per‑transaction limits unless you manually split the work across many chained jobs, which quickly runs into the 50‑job‑per‑24‑hour cap and depth limits.

Why Batch Apex Wins

• Designed for bulk data: Database.executeBatch accepts a QueryLocator that can scope millions of records.
• Each execute call runs in its own transaction, so governor limits reset for every chunk.
• Callouts, DML, and CPU usage are isolated per chunk, reducing the risk that one failure aborts the entire run.
• Batch Apex automatically manages the queue of jobs; you only pay for the number of chunks you need.

Tuning batchSize

Start with Salesforce’s default of 200. For 200k records, a batchSize of 2000 yields 100 chunks, which is usually safe. Increase the size in increments of 500–1000 while monitoring Limits.getQueries(), Limits.getDmlRows(), and Limits.getCpuTime() in a test sandbox. If any chunk approaches the limits (e.g., Limits.getDmlRows() > 10,000), reduce the size. A typical safe range is 500–2000, depending on the complexity of the logic and the presence of callouts.

Batch Size	Chunks (200k records)	Typical Limits
200	1000	CPU 10 s, DML 10 000
500	400	CPU 8 s, DML 10 000
1000	200	CPU 6 s, DML 10 000
2000	100	CPU 4 s, DML 10 000

Impact of Queueable Job Chaining

Queueable jobs can be chained via System.enqueueJob, but:

• Each chained job counts toward the org’s 50 queued jobs per 24‑hour limit.
• Depth is limited to 5 (or 2 in some contexts); exceeding it throws an uncaught LimitException.
• To cover 200k records with a batchSize of 200, you’d need 1000 jobs, far beyond the 50‑job cap.

Practical Steps

1. Create a Database.Batchable class that queries the 200k records via a QueryLocator.
2. Set batchSize to 1000–2000 and test in a sandbox, logging Limits.getQueries() and Limits.getDmlRows() per chunk.
3. If the job completes without hitting limits, deploy to production.
4. Monitor AsyncApexJob after each run to confirm that the total number of jobs stays well below the daily limit.

Ask for Missing Detail

Do you need to preserve state across chunks (e.g., aggregate totals) or perform callouts that must be batched? If so, Database.Stateful and careful callout handling are required.