See why a request is slow, down to the line of code.

Spanrow traces every request across your services and shows which span, query or loop made it slow.

Book a demo

→

See how it works

→

Fig. 1 A checkout request, traced

Live demo, try it

POST/v2/checkouttrace 7f3a91c2e04b1,284ms

Span

200

400

600

800

1.0s

1.2s

POST /v2/checkout

1,284 ms

auth.verify_sessionauth

18 ms

flags.evaluateflags

4 ms

cart.loadcart

41 ms

GET cart:{id}redis

3 ms

SELECT cart_itemspostgres

31 ms

▸ pricing.quotepricing

902 ms

inventory.reserveinventory

94 ms

tax.calculatetax-api

61 ms

risk.scorerisk

58 ms

payments.authorizestripe

168 ms

INSERT orderspostgres

29 ms

publish order.creatednats

7 ms

Last hour, 4,212 checkouts

This one is slower than 94%

median 754 ms

ServiceDatabaseCache, queueExternal call

p95 latency, last 24 h/v2/checkout

1.31sSLO 500 ms · breached

Where this request spent its time

pricing902 ms

stripe168 ms

inventory94 ms

tax-api61 ms

risk58 ms

cart41 ms

postgres29 ms

1Hover a span2Open the slow one3Ship the fix

Three steps from install to root cause.

Install the agent

Add one package and start your server with it. Spanrow finds your web framework, database clients and HTTP calls on its own.

npm install @spanrow/node

node -r @spanrow/node/auto server.js

Send real traffic

Every request becomes a trace: one row per step, timed to the microsecond, across each service it touches.

storefront-api 14 spans

pricing 24 spans

payments 3 spans

Open the slow one

Spanrow ranks requests by the time they cost you and names the span, the query and the line behind each one.

pricing/quote.ts:42

N+1 query, 23 calls

Fig. 2 p95 for POST /v2/checkout, before and after one deploy

Before 4f1c9a2

1.31 s

After 4f1c9a2

0.44 s

Target

0.50 s

Every chart knows what you shipped.

Deploys show up as markers on your latency charts. When p95 moves, Spanrow names the release that moved it and links to the diff.

Read the changelog →

“We had been guessing at checkout latency for a year. Spanrow showed us the loop in pricing on the first afternoon.”

Maya Okafor, Staff Engineer at Halden

What Spanrow records

Everything below arrives from the one package in step 1.

Spans

Every call between services, with start time, duration and status.

Queries

The statement, row count and time of each database call, with values removed.

Logs

Log lines written during a span, attached to that span.

Deploys

Commit, author and time of each release, drawn on every chart.

Targets

A p95 goal for each route, and how much of this month’s budget is left.

Retention

30 days of full traces on Team, 90 days on Enterprise.

Trace your slowest endpoint this afternoon.

Book a demo

→

Free for one service. No card needed.