Speed Wins: From Kafka to APIs in Minutes

@yourtwitterhandle | developer.confluent.io
What are the best practices to debug client applications
(producers/consumers in general but also Kafka Streams
applications)?
Starting soon…
STARTING SOOOOON..
Starting sooooon ..
Starting soon…
Starting soon…

applications)?
Starting soon…
STARTING SOOOOON..
Starting sooooon ..
Starting soon…

Copyright 2021, Conﬂuent, Inc. All rights reserved. This document may not be reproduced in any manner without the express written permission of Conﬂuent, Inc.
Streaming Architecture

applications)?

Goal
Partners Tech Talks are webinars where subject matter experts from a Partner talk about a
specific use case or project. The goal of Tech Talks is to provide best practices and
applications insights, along with inspiration, and help you stay up to date about innovations
in confluent ecosystem.

Starting soon…
STARTING SOOOOON..
Starting sooooon ..

Conﬂuent Perspective Real Time Analytics
9

10
Business challenges Technical challenges
Increasingly complex application
environments and mounting pressures to
track and respond to every indicator and
issue.
Data latency and lack of end-to-end,
scalable observability for monitoring
behavior, performance, and health of
complex systems, applications, and
infrastructure.
Technology failures and security risks
that result in disruption to
customer-facing services and costly losses
for the business.
Higher operational costs due to more
troubleshooting time, bottlenecks, and
suboptimal performance requiring
additional resources/infrastructure.
INDUSTRY: ALL

11
Why Conﬂuent
Stream
data everywhere, on premises and in every
major public cloud.
Connect
operational data like logs, metrics, and traces
from across your entire business including
on-prem, cloud, and hybrid environments.
Process
data streams to feed real-time analytics
applications that query and visualize critical
metrics at scale including latencies, error
rates, overall service health statuses, etc.
Govern
data to ensure quality, security, and
compliance while enabling teams to discover
and leverage existing data products.
Business impact
Enable early detection of system-wide issues
to prevent incidents and downtime.
Deliver proactive, faster responses to open
incidents for quicker resolution.
Gain the ability to deeply analyze all systems
and make more informed decisions.
INDUSTRY: ALL

Why?
Connect
Connect natively to Conﬂuent
and develop scalable APIs in
minutes with SQL.
Share
Share as high-concurrency,
low-latency APIs with other
engineers so they can start
building.
Combine
Combine Kafka data with
other sources (Snowﬂake,
BigQuery, etc.) to build rich and
fast data products.

Speed Wins.
Data Platforms must be fast.
Engineers must also be fast.

Tinybird + Confluent
From Confluent to APIs in minutes.
Unify streams, files,
tables, and more
Develop faster with SQL
and publish APIs
Empower others to build
data products
Real-time personalisation
User-facing analytics
Operational intelligence
Streams
Files
DB tables
Data sources Data products
API Endpoints
Milliseconds

Examples of user-facing analytics
Build differentiated features that delight users
In-product dashboards Real-time personalisation Fraud + anomaly detection
Capture data about user interactions within an application, send that data to an
analytics platform, and build metrics that are then served back to the user as
dashboards or features.

Status quo
Data
Warehouse
Data
Modeling
ETL
Low-Latency
Store
Backend
Data
app
Frontend
Confluent
Common data stack for analytics

New approach
Data
Warehouse
Data
Modeling
ETL
Low-Latency
Store
Backend
Data
app
Real-time data
platform
Frontend
The real-time way
Confluent

Example
User-facing Analytics Architecture

About Tinybird
We accelerate data
and engineering
teams
➔ Open-source ClickHouse at the core
➔ Serverless & fully managed
➔ Cloud-native
➔ Consumption based
➔ Unrivaled developer productivity

"Real-time data is the new
standard, and we want to win.
The best way to deliver a
differentiated user experience is
with live, fresh data."
Damian Grech, Director of Engineering, Data Platform

FanDuel relies on
Tinybird for real-time
personalization and
observability
Processed per month.
273TB+
Requests per day.
Average query latency.
216K+
<50ms
Development time for
the first use case.
<3w

Retail operates in
real-time with
Tinybird
Rows read during
Black Friday
11.9T
Internal Users
+1000
➔ Real-time business intelligence
➔ Real-time inventory management
➔ Real-time personalization
➔ Real-time in-house Web Analytics
P95 latency
240ms
With numbers
Top 5 Global Fashion Retailer

Canva relies on
Tinybird to deliver
insights to their
users
Peak API
1250 rps
➔ Real-time ingestion and analysis of
web events
➔ Real-time User-facing insights about
users published videos
With numbers

Split calculates
the impact of A/B
tests in real-time
Avg. ingested from Kafka
per month (compressed).
220TB+
Requests per day.
From 30-min latency to
real-time.
2.5M+
1-3s
features in production
within 4 months of signing
7
IMAGE

Factorial built 12
new product
features in 6
months
Processed per month.
65TB+
Requests per month.
Average feature dev time
1m
2 weeks
Initial POC to Production
launch time
1 month
IMAGE

The Hotels Network
provides real-time
competitive insights
and personalized
booking experiences
➔ User-facing dashboards and real-time
personalization.
➔ Streaming join in ksqlDB.
➔ Ingested into Tinybird for historical
enrichment and publication via API
Endpoints with sub-second latency.
API requests per month
1B+
Processed per month
6PB

Working together
The ideal joint Confluent
+ Tinybird customer
Performance is table stakes. Tinybird +
Confluent enable engineering teams to develop
faster and ship more..
Are they trying to adapt their existing DW?
Implement a new database? Simplify their stack?
Tinybird + Confluent are better in the cloud.
Prioritise speed to market
Moving from batch to real-time DSP
Cloud native

Thank you.
Confluent Partners do more with Tinybird

Speed Wins: From Kafka to APIs in Minutes

Recommended

More Related Content

Similar to Speed Wins: From Kafka to APIs in Minutes (20)

More from confluent (20)

Recently uploaded (20)

Speed Wins: From Kafka to APIs in Minutes