Phase 2: Operation Test Suites

Goal

Create three operation-measurement test suites so TendSocial can measure the Prisma database cost and performance impact of each API route, standard user flow, page load, click, and action.

Prisma pricing is based on database operations. Prisma states that it counts Prisma ORM queries, not every SQL statement generated behind the scenes. Source: Prisma pricing.

Metrics to capture

Each measured action should capture:

Metric	Meaning
measurementId	Unique UUID for request/action correlation.
label	Human-readable action label.
apiDurationMs	Backend request duration.
browserDurationMs	User-visible browser action duration.
prismaOrmOps	Count of Prisma ORM operations.
sqlStatements	SQL statement count generated by Prisma.
sqlDurationMs	Total SQL duration from Prisma query events.
networkRequests	API calls triggered by a browser action.
rowsRead	Optional if available.
rowsWritten	Optional if available.
warnings	N+1, duplicate fetches, slow query, missing-index suspicion.

Instrumentation foundation

Prisma query logging

Prisma Client supports event-based logging with $on('query'), including query, params, and duration. Source: Prisma Client logging.

Example:

import { PrismaClient } from "@prisma/client";

export const prisma = new PrismaClient({
  log: [
    { emit: "event", level: "query" },
    { emit: "stdout", level: "error" },
    { emit: "stdout", level: "warn" },
  ],
});

prisma.$on("query", (event) => {
  console.log({
    query: event.query,
    params: event.params,
    durationMs: event.duration,
  });
});

Request correlation

Every API request and Playwright action should include:

x-tendsocial-measurement-id: <uuid>
x-tendsocial-measurement-label: dashboard.load

Backend responses should include headers where possible:

x-db-orm-ops: 8
x-db-sql-statements: 11
x-db-sql-duration-ms: 42
x-api-duration-ms: 123

And optionally a JSON debug payload in benchmark mode:

{
  "measurementId": "...",
  "label": "dashboard.load",
  "apiDurationMs": 123,
  "prismaOrmOps": 8,
  "sqlStatements": 11,
  "sqlDurationMs": 42
}

Suite A: Backend API operation tests

Goal

Call each API route directly from the backend test runner and measure the Prisma impact of every route without browser noise.

Command

pnpm test:dbops:api

Coverage checklist

API area	Required actions
Auth/session	current user, session check, workspace membership
Workspace	list, load, switch, update
Brand profile	load, update, brand voice/tone
Content	list, detail, create, update, archive/delete
Blog	list, load body, save draft
Calendar	month load, week load, schedule item
Campaigns	list, detail, attach content
Social accounts	list, connect placeholder, sync metadata
AI workflow metadata	create draft, save generation result, update status
Publishing queue	enqueue, list, update status
Notifications	list, mark read
Settings	load/update account and workspace settings

Output

reports/dbops/api-summary.json
reports/dbops/api-summary.csv
reports/dbops/api-summary.md

API result row

{
  "endpoint": "/api/content",
  "method": "GET",
  "actionLabel": "content.list",
  "status": 200,
  "apiDurationMs": 118,
  "prismaOrmOps": 7,
  "sqlStatements": 9,
  "sqlDurationMs": 39,
  "warnings": []
}

Suite B: Playwright standard task run

Goal

Measure representative user workflows through a real browser.

Playwright supports reusable authenticated browser state, which avoids logging in during every test. It recommends storing auth state under playwright/.auth and keeping it out of source control because it can contain sensitive cookies/headers. Source: Playwright authentication.

Command

pnpm test:dbops:e2e-standard

Standard flows

Flow	Steps
App shell	Open app, load dashboard, load workspace
Content creation	Create social post, save draft, view list
Blog draft	Create blog post, edit body, save
Calendar scheduling	Open calendar, schedule content, change date
Brand settings	Load brand profile, update voice/tone settings
Social accounts	Load connections, open account detail
AI draft generation	Trigger mocked AI generation, save result
Publishing queue	Enqueue mocked post, view queue/status

Playwright network tracking

Playwright can monitor request and response events and wait for network responses after user actions. Source: Playwright network.

Use:

page.on("request", request => {
  // capture request.method(), request.url(), headers
});

page.on("response", async response => {
  // capture response.status(), response.url(), DB measurement headers
});

Output

reports/dbops/e2e-standard.json
reports/dbops/e2e-standard.csv
reports/dbops/e2e-standard.md

Suite C: Exhaustive Playwright action/page-load crawl

Goal

Measure every page load, click, modal open, tab switch, filter, search, sort, pagination, save, and destructive action in the app.

This is a DB profiler, not a QA suite.

Command

pnpm test:dbops:e2e-exhaustive

Action registry

Create:

tests/dbops/action-registry.ts

Example:

export const actions = [
  {
    id: "dashboard.load",
    route: "/dashboard",
    steps: [
      { type: "goto", url: "/dashboard" }
    ],
  },
  {
    id: "content.open-create-modal",
    route: "/content",
    steps: [
      { type: "goto", url: "/content" },
      { type: "click", role: "button", name: "Create" }
    ],
  },
  {
    id: "content.filter.status-draft",
    route: "/content",
    steps: [
      { type: "goto", url: "/content" },
      { type: "click", role: "button", name: "Status" },
      { type: "click", role: "option", name: "Draft" }
    ],
  }
];

Measurement wrapper

For every action:

const measurementId = crypto.randomUUID();

await page.setExtraHTTPHeaders({
  "x-tendsocial-measurement-id": measurementId,
  "x-tendsocial-measurement-label": action.id,
});

Collect all API responses and sum DB measurement headers.

Output

reports/dbops/e2e-exhaustive.json
reports/dbops/e2e-exhaustive.csv
reports/dbops/e2e-exhaustive.md
reports/dbops/e2e-exhaustive-hotspots.md

Hotspot ranking:

highest Prisma ORM operations
highest SQL statement count
highest SQL duration
highest browser duration
highest repeated API calls
highest background/polling noise

Traces

Enable Playwright trace on failure or retry. Trace viewer is useful for debugging which browser actions and network requests created a DB hotspot. Source: Playwright trace viewer.

Suggested config:

use: {
  trace: "on-first-retry",
}

For dedicated investigation:

npx playwright test --trace on

Initial thresholds

Action class	Target Prisma ops
Simple read	1-3
Workspace/dashboard shell	<10
Calendar/content list	<15
Heavy editor save	<10
AI generation save	<10
Background batch job	Bounded and paginated

Acceptance criteria

Item	Target
API suite created	Yes
Standard Playwright suite created	Yes
Exhaustive Playwright suite created	Yes
Reports emitted as JSON, CSV, and Markdown	Yes
Measurement IDs correlate browser/API/Prisma activity	Yes
Top 20 DB hotspots identified	Yes
CI can run API suite	Yes
Exhaustive suite can run manually/on demand	Yes

Sources

• Prisma pricing
• Prisma Client logging
• Playwright authentication
• Playwright network
• Playwright trace viewer