One data layer for every database and file
CKN Data is the framework's data-access layer: six database types, Excel, Word, CSV files and data pipelines behind one consistent API. Your team learns it once, moves between systems easily, and avoids a pile of third-party libraries to keep patched.
What CKN Data does
Every database, one way of working
MySQL, PostgreSQL, Oracle, SQL Server, MongoDB and Firestore share the same commands. Switching databases means minimal code change, and drivers load only when a connection is actually opened.
- query, insert, update, delete, count, upsert
- Check each database's abilities with capabilities
- Configurable connection pool on every pooled adapter
- Parameterized queries guard against SQL injection
Transactions that are safe by default
Any call made inside a transaction block automatically joins it, even through other functions. If anything fails, everything rolls back.
- Nested transactions through savepoints
- Independent transactions, e.g. to log why a request was refused
- Row locks with SELECT ... FOR UPDATE to prevent write races
- MongoDB retries transient write conflicts automatically
Big data volumes without memory spikes
Read hundreds of thousands of rows as a stream, one batch at a time, and write in bulk with batch sizes tuned for each database.
- stream() on every adapter except Firestore
- queryColumnFrame() for aggregating large results
- Multi-row upsert and bulkUpdate in one statement
- Per-statement timeouts
Excel, Word, PowerPoint, CSV and ZIP
Create, read and edit office documents straight from code: reports, quotations, and imports from files your customers send.
- ExcelFile, WordFile, PowerPointFile and Excel Online
- Extract document text for an AI model to read
- Atomic writes and folder confinement for files
- Metered ZIP opening to guard against hostile archives
Pipelines and connectors to other systems
Build ETL/ELT jobs that move data between databases, files and message brokers, with checkpoints to resume a run that failed midway.
- HTTP, S3, SFTP, Kafka, Redis Streams, CDC, DuckDB, SOAP
- Checkpoints and dead-letter storage for records that fail
- Late records are counted and set aside, never dropped silently
- Declarative pipeline specs
The data housekeeping real systems need
Includes the modules business systems depend on, from guaranteed event delivery to schema upgrades and Thai-aware text search.
- Outbox: safe retries without duplicates, via idempotency keys
- Schema migration planning with column-type checks before running
- Validation, import/export and data sync
- RAG storage on PostgreSQL with Thai-aware full-text search
Measured, not promised
Figures come from our own benchmark harness, run on an Apple M1 Pro with Docker (database containers capped at 2 CPUs and 2 GB, 100,000 seeded rows). Absolute values on a laptop differ from production servers; the ratios against the raw driver and before/after comparisons are the useful part.
Figures come from CKN's own test suites and benchmarks. Real results depend on each system's hardware and data volume.
What your business gets
Lower build and maintenance cost
Your team uses one API for databases, files and pipelines, so new developers get productive faster and code written once is reused across systems. Jobs that normally need extra libraries, such as producing Excel or Word files, are covered in the same package.
Less risk where money and stock are involved
Full rollback, row locking and an Outbox that retries without duplicating events help prevent problems like negative stock or double orders. The test suite also runs against real servers for all six databases, not just mocks.
You own the stack, with no single-vendor lock-in
Word writing, CSV and XML handling are our own code, which cuts exposure to third-party libraries you would have to keep patching. You install only the drivers for the databases you use, and you can move to another database without rewriting the whole data layer.
Example: load prices, then deduct stock without a race
In use across the CKN stack
- All 14 CKN business services use it as their main database layer (PostgreSQL)
- Platform services such as identity, integration, portal, intelligent and engine
- The CKN Works internal app (projects, tickets, costing, documents, expenses)
- The ckn command-line tool generates database-layer code for new apps
import { PostgreSQL } from "@ckn-technology/data/database";
const db = new PostgreSQL({
host: process.env.MAIN_DB_HOST ?? "", port: 5432, user: process.env.MAIN_DB_USER ?? "",
password: process.env.MAIN_DB_PASSWORD ?? "", database: "app", schema: "sales", enableLog: false,
});
await db.upsert("price", [{ sku: "A-100", company_code: "TH01", price: 890, currency: "THB" }], ["sku", "company_code"]);
await db.transaction(async (tx) => {
const [s] = (await tx.query("SELECT qty FROM stock WHERE sku = $1 FOR UPDATE", ["A-100"])).rows;
if (!s || s.qty < 5) throw new Error("not enough A-100"); // rolls everything back
await tx.update("stock", { qty: s.qty - 5 }, "sku = $1", ["A-100"]);
});
await db.close();