Tobias Lins
@tobiaslins
Product Engineer
· Building @Vercel Observability Products
· Splitbee Founder (acquired by Vercel)
748 Following 6.5K Followers
Hackathon project from @tomlienard and myself - so happy I can finally measure all the things
Emit custom metrics from Vercel Functions and query them in Observability alongside Vercel's built-in data.
Call 𝚖𝚎𝚝𝚛𝚒𝚌() from @𝚟𝚎𝚛𝚌𝚎𝚕/𝚏𝚞𝚗𝚌𝚝𝚒𝚘𝚗𝚜 with a name, a value, and any attributes you want to filter and group by.
Show more
The beginning
We've been using similar concepts when building Vercel Data Pipeline
- Processing over 5GB/s
- Supports multi fanout
- Deduplication
- At-least once delivery
I think I need to write a blogpost
I'm pretty sure the only reason to use Kafka for telemetry data in 2026 is muscle memory.
A simple, open source, S3-backed pipeline can shuttle 1Gb/s of logs to @ClickHouseDB for $200/mo with 2.8s p50 latency. Fast enough for us.
Show more
Building a new high performance chart library
This has 120fps while rendering 15m datapoints
- Lightweight
- Can be server rendered
- Mobile optimized
... much more
You can just ship Postgres interface support for your favorite upcoming relational db
We've migrated our data pipeline to use Vercel Queues.
Each queue item is just a lightweight marker that says, "There's a new chunk of data available." The actual data lives in S3.
Using Queues, we can now fan out processing into independent consumer groups, for example:
→ Insert the data into ClickHouse
→ Convert it to Parquet and write it back to S3
→ Sample it and send it elsewhere
Each consumer group is isolated, so if one falls behind or experiences issues, the others continue processing uninterrupted.
Show more
We've been using similar concepts when building Vercel Data Pipeline
- Processing over 5GB/s
- Supports multi fanout
- Deduplication
- At-least once delivery
I think I need to write a blogpost
Hi @GitHubDesktop - Can we please add sorting of branches by last pushed?
It's so hard to find the right one otherwise 🙏
Building a new high performance chart library
This has 120fps while rendering 15m datapoints
- Lightweight
- Can be server rendered
- Mobile optimized
... much more
Looking goooood, @vercel 😎
Yep, you read that right.
You can now run any Dockerfile on Vercel.
# 𝙳𝚘𝚌𝚔𝚎𝚛𝚏𝚒𝚕𝚎.𝚟𝚎𝚛𝚌𝚎𝚕
𝙵𝚁𝙾𝙼 𝚐𝚘𝚕𝚊𝚗𝚐:𝟷.𝟸𝟺
𝙲𝙾𝙿𝚈 . .
𝚁𝚄𝙽 𝚐𝚘 𝚋𝚞𝚒𝚕𝚍 -𝚘 /𝚜𝚎𝚛𝚟𝚎𝚛 .
𝙲𝙼𝙳 ["/𝚜𝚎𝚛𝚟𝚎𝚛"]
Show more
We are so back
WebSockets are now supported on Vercel.
Build realtime apps using standard Node.js libraries, including socket.io.
Introducing eve, an agent framework.
𝚊𝚐𝚎𝚗𝚝/
𝚊𝚐𝚎𝚗𝚝.𝚝𝚜
𝚒𝚗𝚜𝚝𝚛𝚞𝚌𝚝𝚒𝚘𝚗𝚜.𝚖𝚍
𝚝𝚘𝚘𝚕𝚜/
𝚜𝚔𝚒𝚕𝚕𝚜/
𝚜𝚊𝚗𝚍𝚋𝚘𝚡/
𝚜𝚌𝚑𝚎𝚍𝚞𝚕𝚎𝚜/
Like Next.js, for agents.
Show more
We've been using similar concepts when building Vercel Data Pipeline
- Processing over 5GB/s
- Supports multi fanout
- Deduplication
- At-least once delivery
I think I need to write a blogpost
I'm pretty sure the only reason to use Kafka for telemetry data in 2026 is muscle memory.
A simple, open source, S3-backed pipeline can shuttle 1Gb/s of logs to @ClickHouseDB for $200/mo with 2.8s p50 latency. Fast enough for us.
Show more
@ everyone who has a big data pipeline (logs, metrics, OTEL)
- What are you using? Kafka, Kinesis, ...?
- Are you happy with it?
- Do you feel like its too expensive?
I'd love to talk!
The „Vercel Data Format“ got born.
I hope I can share more about this soon
I love building observability pipelines
At @vercel scale completely new problems arise compared to our 1B events/month at Splitbee
→ A single customer can have billions of datapoints week
→ Data streams are at GB/s scale
→ Achieve e2e latency of <5s
→ Durability is key. We should never loose data when the database is down
Show more
This is just such a wonderful way to define drip campaigns.
Wait 2 days to send a reminder, but don't do it when a user does X.
@workflowsdk's Hook primitive lets external events control a running workflow.
Let's take the email drip sequence example - welcome email, then nudges at day 2 and day 4. But what if the user completes the action before the nudges fire? You don't want to keep nagging them.
The pattern you can use to solve this: Define a hook, then race it against each sleep.
If the hook fires first, skip the remaining nudges. Any service in your app can resume the hook - it just needs the user's email as the token.
Show more
I got the chance to see it early and it's 🤯
Every day I'm praying so you can get it into your hands soon
we’ve been a bit quiet…
it will all make sense very soon 🤫
I love building observability pipelines
At @vercel scale completely new problems arise compared to our 1B events/month at Splitbee
→ A single customer can have billions of datapoints week
→ Data streams are at GB/s scale
→ Achieve e2e latency of <5s
→ Durability is key. We should never loose data when the database is down
Show more
