End-to-End Booking Flow

Step-by-step walkthrough

Read path — search and browse

• ① Browse / filter — Guest runs date, price, and location filters against the Search API.
• ② Query index — Search API queries Elasticsearch inverted index — never scans inventory rows.
• ③ Cache hit — Hot listing cards served from Redis when the same queries repeat.
• ④ GET listing detail — Guest opens a listing page; photos and description from search index or cache.

Write path — hold, pay, confirm

• ⑤ POST /hold — Guest reserves a slot; Booking API starts a short transaction.
• ⑥ Lock slot — Inventory row locked in PostgreSQL (SELECT FOR UPDATE or serializable isolation).
• ⑦ POST /confirm — Guest submits payment with Idempotency-Key to prevent double charge on retry.
• ⑧ Charge card — Booking API calls Stripe; payment success required before confirm.
• ⑨ Confirm → COMMIT — Hold converted to sale in one atomic transaction.
• ⑩ booking.confirmed — Event published to Kafka after commit (outbox pattern).
• ⑪ Async index update — Consumer updates Elasticsearch availability — derived data, not dual-write.

// Hold → pay → confirm with compensation
async function bookingSaga(bookingId: string) {
  const hold = await inventory.hold(bookingId);
  try {
    const charge = await stripe.charge({ idempotencyKey: bookingId });
    await inventory.confirm(hold.id);
    await kafka.publish("booking.confirmed", { bookingId, chargeId: charge.id });
  } catch (err) {
    await inventory.release(hold.id); // compensating action
    throw err;
  }
}