Modernize Data Platform with
SingleStore
—
Agenda
—
● Introduction to SingleStore
Nguyen Thanh My
● What makes SingleStoreDB unique
Solution Engineer
● Demo: Real Time Marketing Analytics
● Summary
● Q&A
SingleStoreDB is the cloud-native database built with
speed and scale to power data-intensive applications
3
SingleStore - Growth & Momentum
FINANCIAL STRENGTH CUSTOMER & REVENUE
— COMMUNITY EMPLOYEES
$ 278M 209% 1M+ 400+
Raised in the last 20 YoY Cloud Customer Downloads of free Employees in 9 countries
months alone growth production SingleStoreDB Growing at 24% YoY
Fiscal ‘22 ending February 2022
5.0 5.0
“Best SQL distributed database “I don't think you can beat
available in 2022” SingleStore. It’s a ‘Ferrari’ of a DB”
June 13, 2022
April 5, 2022
IN 4 CATEGORIES
Recognized in Gartner MQ 5.0 10 out of 10
for Cloud DBMS -2021
“SingleStore - the only database “SingleStore is the fastest
you will ever need” database ever tested by us”
June 23, 2021 August 18, 2021
Modern IoT
Applications are Dynamic Wealth
everywhere Cybersecurity Management
— Dynamic Customer
● Access to real-time data Experiences
● Deliver fast and highly interactive
customer experiences
● Scale effortlessly Supply Chain Real-Time
Analytics Marketing Analytics
● Run anywhere
Anomaly Detection
5
Data Intensive Applications
3 Key Required Capabilities
Ultra-Fast Ingest Super-Low Latency High Concurrency
Parallel, high-scale Blazing Fast Unparalleled
streaming data Ingest Queries Scalability
Billions of events/sec for Sub-second latencies with Millions of real-time queries across
immediate availability immediate consistency tens of thousands of users
Data Intensive Apps
Fast analytics on dynamic data
for complex analytical queries
6
BUILDING A DATA-INTENSIVE APPLICATION
Solving Data Intensity Through Complexity
—
Application
3rd Party
Cache Analytics
“My applications “The data in our
“We can’t scale fast “I don’t want to
aren’t fast or interactive application is not fresh!
enough! We are running move the data again!
enough. It’s starting to We can only get data
into data bottlenecks. Want to minimize AI / the
ML
impact the customer
Operational from day(s) ago or
Its impacting my SLAs” complexity (and costs)!”
experience!” from few hours ago!”
Batch Ingest
Batch
Search
Streaming Data
IoT/ Clickstream
7
BUILDING A DATA-INTENSIVE APPLICATION
Solving For Data Intensity The Modern Way
—
Application
3rd Party
Transactions and
Analytics
Operational
AI / ML
Batch
IoT
8
SingleStoreDB
—
Performance, Scale & Concurrency
● Process more than 1 trillion rows per second
● Parallel streaming ingestion of millions of events/sec
● Low-latency, single-digit millisecond responses
● Scale client concurrency from 1 to 1000s
Price
● 60% better price-performance & TCO
over using a combination of databases
● 10-100x the performance at ⅓ the cost
over legacy databases
Simplicity
● Familiar SQL and MySQL wire-protocol compatibility
● Multi-Model - relational, document, time-series, full text
search, geospatial, key-value, vector SingleStoreDB is a distributed relational,
● Cloud, On-prem, containers, bare metal multi-model cloud database built for real-time
applications and analytics
9
$10M+
Additional annual revenues
and 500% cost reduction
Billing processes from monthly to daily
3500+
Users Supported
Real-time insights with massive
concurrency for marketing and analytics
50% $40M $2B
Reduction in data Annual operational 2 Power Plants worth
infrastructure costs cost savings of electricity saved annually
Streaming analytics for proactive care Multi-channel customer data hub for IoT Analytics to balance electric grids
and real-time recommendations 120M US customers at reduced TCO in real time
Why
—
Customers Adopt SingleStore?
CUSTOMER EXP FASTER INSIGHTS LOWER TCO GROW FASTER
●“My applications aren’t fast or ●We need insights faster!
●My cloud-costs are too high ●I can’t scale fast enough!
●Need to see the data in minutes
Business responsive enough”
- not hours”
●I’m tying together multiple ●I’m not getting the performance
●Its impacting the customer exp. databases for my application as we scale
Challenges ●Our dashboards are too slow
●It’s way too complex ●It’s way too complex and costly
●I can’t scale fast enough! and not interactive
—How
SingleStore
Helps
Migrate from legacy
FastApps Fast Analytics Database Sprawl
data stores
●Drive new revenue streams ●Make informed business ●Reduce operating costs ●Scale your business fast
Business ●Deliver an engaging customer decisions in the moment! ●Remove perf.bottlenecks ●Reduce complexity and cost
experience ●Improve time-to-insights
Value ●Accelerate your growth ●Real-time customer experience
●Simplify the data landscape ●Drive a competitive edge
11
GigaOm Benchmarking Study - Summary
—
Source: GigaOm Report on
Transactional and Analytical
Workloads, June 2022
Key Highlights
50% Savings 60% Savings Up to 100% Faster
over three years compared over three years compared in TPC-H workloads compared
to Snowflake-MySQL stack to Redshift-PostgreSQL stack to Redshift (with Refresh)
12 Return to menu
Link: https://www.singlestore.com/resources/report-transactional-and-analytical-workloads-a-gigaom-database-benchmark/
Adopting SingleStore
Augment Modernize Build
Augment DBs or Analytical DWs Consolidate and Replace Build new modern applications
Augment legacy and speciality Modernize by consolidating/ replacing Net new apps or operational
OLTP/ OLAP engines, or Hadoop legacy datastores, speciality databases, or analytics workloads
Hadoop
13
What makes SingleStoreDB unique?
—
Unified: Transactions Streaming Ingestion
+ Analytics & Analytics
The only database with a single, Ultra-fast ingestion with
unified table type for both non-blocking low-latency
Transactions & Analytics analytics on data as it lands
Price - Performance Unlimited Storage
World's only database that World's only fast operational
delivers best-in-class performance database with separation of
on TPC-C, TPC-H and TPC-DS storage and compute
Transactions + Analytics
14
Unique Capabilities
—
15
Fast Data Ingest - SingleStore Pipelines
—
Product Overview
● Streaming ingestion with immediate availability
● Fast parallel bulk loading - millions of events/sec
● Built-in Create Pipeline
● Transactional Consistency
● Exactly-Once Semantics
● Native integrations with Kafka, Cloud Storage,
AWS S3, Azure Blob, HDFS
ENGINEERING
How to Load 100 Billion Rows
of Data Ultrafast
Nabil Nawaz
16 How to Load 100 Billion Rows of Data Ultrafast, SingleStore Blog
Anatomy of Pipelines
—
Write to Kafka
Hadoop
Distributed File System
Google Cloud Storage
Kafka
Microsoft Azure
Blob Storage
Amazon S3
SingleStore pulls for Data can be directly inserted
changes from a data into tables or pre-processed by a
source system. stored procedure.
SingleStore pulls the data into its The data is committed in a
memory space (no commit) where a transaction (and in parallel).
transform can be applied.
17 Return to menu
Universal Storage
— U.S. patent number
11,068,454
Old: Two Table Types New: Universal Storage
(Most RDBMS do this) (Single Table Type for Transactions & Analytics)
Capabilities
Rowstore
● Allows fast seeks and analytics
● Great transactional workloads
● Support fast insert/ update/
● Not great for analytics
delete/ upsert
● Support integrity constraints
Benefits
Columnstore
● Simplifies data architecture
● Great for analytics
● Faster apps (better CX)
● Can't do many
transactional workloads ● Reduced data movement,
duplication and hardware costs
18 Return to menu
SingleStoreDB Multi-model capabilities
—
Relational Time-Series
Standard SQL plus ability to join Native date, time, and timestamp types
across the multi-model schema with built-in functions, e.g. time_bucket
Key-Value
Unstructured, Full-Text Search
Scan >30B tuple space and return trillions
Built-in, native inverted indexing
of key-value results in < 1 second
GeoSpatial Semi-Structured, Document
Polygons, points, lines and built-in JSON as native type for schema-on-write
spatial functions and schema-on-read
19 Return to menu
Real-Time Marketing
Demo Analytics with
— SingleStoreDB
20
USE CASE OVERVIEW
Demo: Real-Time Marketing
Offers & Analytics
—
Fictional service company with millions of
subscribers serving targeted ads and offers to
customers in real time
Targeted ads based on customer demographics,
geo-location, purchase history etc.
Evaluate 80 Million ad-serving opportunities in
real-time and serving ~30-50K ads/offers in
real-time to customer’s cell phones
21
Data Sources & Data Types
Real-Time & Batch Sources
—
Data Sources
● Purchase history ● Offers
(millions of purchases) (sub-second matching)
● Subscriber data ● Web history
● Streaming subscriber location (time-series data)
(geo data and functions) ● Notifications
Data Types
Relational GeoSpatial JSON Time-Series
22
Demo - Logical Flow
—
Dashboard Ad-Hoc Queries
(RW) (R)
Data 80M Push offers
Sources notification / Ads
opportunities
WS 1 WS 2 Under 500ms
SingleStore
Pipelines
Matching Notifications
Matching Algorithm Location-based
to match offers with Push notifications
Amazon S3 subscribers User History
SingleStoreDB
5M+
notifications
in real-time
Offers Segments
Max bid price Location
Activation Zone User History
Segments
23
Summary
—
24
Why SingleStore?
—
20-100x Up to 60% Lower Real-Time
Performance Boost Costs & Complexity Operational Analytics
Faster Eliminate the need for Bring Real-Time Data
Insights Now! Multiple Data Engines Experience to your Apps
Drive up to 20-100x* performance Multi-model database with scale Distributed SQL database with
improvements compared to Gen-1 out HTAP so that you can get rid transactions and analytics in the same engine
single node databases of speciality data engines with no data movement
*Based on real-life customer examples *Based on real-life customer examples
25
Try SingleStoreDB
Cloud for free
today Get Started with
— $500 in Free Credits Today
Get SingleStoreDB’s best-in-class speed,
scale, and agility without the headaches
of installing, configuring, and maintaining
software. Go to www.singlestore.com/cloud-trial
26
Thank you!
—
27
