Comparisons

VeloDB vs Redshift

While Redshift serves as a cloud data warehouse, VeloDB is engineered for superior real-time analytics, delivering higher concurrency, a flexible architecture, and seamless, non-disruptive scaling.

Start Free Cloud Trial

Faster Query

Higher Concurrency

SecondsvsMintues

Real-Time Data Latency

Multi-Cloud

Why Choose VeloDB

	VeloDB	Redshift
Deployment	Supports multiple deployment options on multi-cloud SaaS and BYOC service on AWS, Azure, and GCP On-premise deployment with enterprise-grade reliability	Supports only SaaS service on AWS
Architecture	Multi-compute, shared-data architecture Compute and storage are fully decoupled Enables resource isolation at the compute group level, and further through workload partitioning within those groups	Lacks multi-cluster shared storage; resource isolation is only achievable via workload partitioning within a single compute group
Real-Time Ingestion & Updates	High-throughput real-time data updates reach millions of records per second Consume data from sources like Flink, Kafka, and APIs in real-time, with data visibility in seconds	Not ideal for frequent data updates Batch data ingestion
Data API	Supports Arrow Flight for high-speed data reading	Only supports low-speed data reading via JDBC, ODBC
High Concurrency	Supports high concurrency for various analytical workloads Uses hybrid row-column storage for efficient point queries	Relies on columnar storage, which limits high-concurrency queries.
Rich Indexes	Skip Index: Minmax Index, BloomFilter Index Point Query Index: Prefix Index, Inverted Index	Only supports skip index (Minmax Index, BloomFilter Index)
High Availability	Cluster scaling (scale-out/in) does not impact ongoing business operations	Cluster scaling (scale-out/in) causes downtime of minutes
Use Cases	Real-Time Analytics Data Warehouse and Lakehouse Logging and Observability	Supports data warehouse and lakehouse, yet not ideal for real-time analytics and observability

VeloDB

Deployment
Supports multiple deployment options on multi-cloud

SaaS and BYOC service on AWS, Azure, and GCP
On-premise deployment with enterprise-grade reliability
Architecture
Multi-compute, shared-data architecture

Compute and storage are fully decoupled
Enables resource isolation at the compute group level, and further through workload partitioning within those groups
Real-Time Ingestion & Updates

High-throughput real-time data updates reach millions of records per second
Consume data from sources like Flink, Kafka, and APIs in real-time, with data visibility in seconds
Data API

Supports Arrow Flight for high-speed data reading
High Concurrency

Supports high concurrency for various analytical workloads
Uses hybrid row-column storage for efficient point queries
Rich Indexes

Skip Index: Minmax Index, BloomFilter Index
Point Query Index: Prefix Index, Inverted Index
High Availability

Cluster scaling (scale-out/in) does not impact ongoing business operations
Use Cases

Real-Time Analytics
Data Warehouse and Lakehouse
Logging and Observability

Redshift

Deployment

Supports only SaaS service on AWS
Architecture

Lacks multi-cluster shared storage; resource isolation is only achievable via workload partitioning within a single compute group
Real-Time Ingestion & Updates

Not ideal for frequent data updates
Batch data ingestion
Data API

Only supports low-speed data reading via JDBC, ODBC
High Concurrency

Relies on columnar storage, which limits high-concurrency queries.
Rich Indexes

Only supports skip index (Minmax Index, BloomFilter Index)
High Availability

Cluster scaling (scale-out/in) causes downtime of minutes
Use Cases

Supports data warehouse and lakehouse, yet not ideal for real-time analytics and observability

Performance Comparison

TPC-H SF100 Benchmark

The TPC-H benchmark with a scale factor of 100 (SF100) is a widely used standard for evaluating database performance. It includes a set of complex SQL queries designed to simulate real-world business intelligence workloads.

For comparison of query performance, models with similar costs were selected:

Redshift 5 * ra3.4xlarge
VeloDB Cloud with 112c configuration

ClickBench

ClickBench is a leading benchmarking tool for evaluating analytical database performance. In this test, it specifically compared VeloDB and Redshift's capabilities in both query execution and S3 data loading.

For data ingestion, both VeloDB and Redshift directly load data stored in S3 within the same VPC.

For queries, the benchmark uses real-world data from a major web analytics platform, focusing on large,flat tables to simulate typical clickstream analysis and structured log scenarios .

For comparison of query performance, models with similar costs were selected:

Redshift 5 * ra3.4xlarge
VeloDB Cloud with 112c configuration

Data Ingestion Performance

These tests primarily focus on evaluating data ingestion latency and throughput performance. The end-to-end ingestion latency was measured from when data was produced into AWS MSK until it became queryable.

For the comparative analysis, both Redshift and VeloDB consumed data in real-time from MSK, using the following configurations:

Redshift 5 * ra3.4xlarge
VeloDB Cloud with 112c configuration