AVE.ai
~1,000 QPS
P99 < 1s
at peak
50%+
lower cost
AVE.ai evaluated ClickHouse and Snowflake for real-time token analytics, then chose VeloDB. It now sustains close to 1,000 QPS with data landing in under 5 seconds.
- Products
- SolutionsUse Cases
Real-Time Analytics
Sub-second dashboards and data products on petabytes of data at any concurrency
Data Warehousing
Sub-second analytics on open lakehouse formats with no vendor lock-in
Observability in the AI Era
The most cost-effective alternative to Elasticsearch observability
Context Engineering
Hybrid search and fresh context for RAG, agents, and LLMs
IndustriesFinancial Services
Real-time analytics and fraud detection for banking, capital markets, and on-chain data
Ad & Media tech
Advertiser-facing reporting, content search, and game telemetry on one engine
Automobile & Transportation
Fleet telemetry, real-time recommendations, and multimodal autonomous driving search
- Docs
- Resources
- Pricing
- Contact us
Industry solutions: financial services
VeloDB for financial services
Banks, trading venues, payment platforms, and crypto businesses run on data that changes by the second: prices, positions, balances, and risk signals. VeloDB, the real-time analytics and search database from the creators of Apache Doris, serves trading analytics, fraud detection, and on-chain analysis in a single engine, built for the query concurrency and data update frequency that financial workloads require.
Banking & payments: Power real-time customer-facing analytics, monitor payment infrastructure end to end, and run risk and log analysis across billions of events at high concurrency.
Capital markets: From quant backtests to real-time P&L, work across decades of market data and millions of trade events per second on one live view of positions and risk.
Fraud detection: Score transactions and trading behavior against aggregates computed on data ingested seconds ago, with the concurrency investigation workflows demand.
Digital assets & crypto: Run on-chain analysis and live trading analytics at exchange scale, sustained through the launch-day traffic spikes DeFi brings.
Submit your inquiries here, and our team will get back to you
Customer Stories
Web3 market analytics
Digital identity & risk
Advance.AI
50%+ lower cost
5x
faster queries
Near real-time
log writes
Advance.AI moved its log system from OpenSearch to VeloDB and unified log analysis with analytics on one engine.
Powering real-time data for 10,000+ teams
Challenges working with real-time financial data
01
Real-time analytics stalling in real-world conditions
When markets move or tokens launch, three things happen at once: query concurrency spikes, every price and balance update rewrites hot rows, and analytics queries join tables that are changing mid-query. General-purpose OLAP engines slow down at the moment the business needs them most.
AVE.ai hit all three limits while evaluating ClickHouse and Snowflake for Web3 market analytics. Read their story
02
Mixed workloads, leading to a complicated stitched-together system
Time-series data lands in one store, point lookups run on an OLTP database, and aggregations get their own analytics tier. Every seam adds data movement, drift between copies, and one more system to operate, and each team ends up with its own version of the truth.
BCEL ran HDFS, YARN, Hive, ZooKeeper, and a metastore to serve its reporting layer before consolidating on VeloDB.
03
Search and analytics workloads require different engines
Logs and text sit in a search engine while metrics sit in an OLAP store, so teams pay to store the data twice, maintain two query languages, and join across systems to answer one question.
Advance.AI ran OpenSearch for logs alongside its analytics stack, at 50% higher cost. Read their story
Why VeloDB for financial services
Performance
Fast where speed matters
Merge-on-write updates keep tables current under heavy writes, so balances, positions, and prices are queryable the moment they change. Point queries reach 30,000 QPS per node, complex joins run distributed across the cluster, and high-concurrency dashboards stay responsive through peaks. ASOF JOIN, available in Doris 4.1, pairs each trade with the latest quote for execution analytics and trade surveillance.
Unified
Time-series, OLAP, point lookups and more in one system
One engine handles streaming market data, aggregations over any window, and OLTP-style entity lookups, so analysts share a single up-to-the-second view instead of reconciling per-team stores. Native CDC ingests from PostgreSQL, removing the need for additional middleware.
The next use case is a new index or a new table, not a new database.
Search + Analytics
Search and analytics together
VeloDB provides native support for inverted indexes, so SQL aggregations, log search, entity lookup, and metric analysis run on one copy of the data with one engine and one SQL statement. Doris 4.1 adds BM25 scoring and Elasticsearch-compatible search syntax.
Deploy Anywhere
Runs where regulators expect it to
Deploy as fully managed SaaS, BYOC in your own account, or on-premises for full control over data residency. Storage-compute separation scales for market peaks without paying peak capacity all year. VeloDB also integrates with open table formats such as Iceberg and Paimon, as well as open catalogs such as Apache Polaris and Unity Catalog.
Workloads financial teams run with VeloDB
Banking and regulatory reporting
- Teller and management reporting reconciled across every branch at end of shift
- Hundreds of concurrent report consumers on PowerBI, Tableau, Metabase, and Superset
- Account and transaction history lookups served next to aggregate reporting
- Change data capture from MariaDB, MySQL, and Oracle core banking into one reporting layer
BCEL delivers roughly 1,000 teller and management reports a day on VeloDB, with the analytics team's exploratory querying running on the same data, on a cluster still under 30% utilization.
BCELReal-time trading analytics
- Live P&L and positions refreshed as trades execute
- Strategy dashboards shared across a full analyst pool
- Time-window aggregations over 1-minute, 5-minute, 1-hour, and 24-hour ranges
- Trade-to-quote joins for execution analytics, powered by ASOF JOIN
Transaction fraud detection
- Behavioral scoring against aggregates computed on fresh data
- Anomaly detection for volume spikes and large trades as they happen
- Entity-history lookups inside live decisioning flows
AVE.ai runs trading-behavior analytics, volume, large trades, and fraud detection, on the same engine that serves its market analytics. Read here
AVE.aiOn-chain analysis
- High-throughput ingestion from major blockchains, with heavy updates as chain state changes
- Token, wallet, and pool analytics under launch-day concurrency
- Sub-5-second latency end to end, sustained at AVE.ai
A top crypto exchange serves 80M+ users on VeloDB at 5,000 peak QPS with P95 latency under 500ms. Story publishing soon.
Crypto exchangeFinTech operations and log analysis
- Log search and analytics on one copy of the data
- Near real-time log writes for time-critical business functions
- Fewer systems and a simpler architecture to run
Advance.AI unified its log system on VeloDB and cut costs by more than 50%. Read here
Advance.AIFAQ
Frequently asked questions
What is VeloDB?
VeloDB is a real-time analytics and search database from the creators of Apache Doris. One engine handles point lookups, complex joins, high-concurrency aggregations, and full-text and vector search on data that changes by the second.
How does VeloDB support real-time fraud detection?
Fraud scoring needs fresh aggregates and fast entity lookups at the same time. VeloDB ingests transaction streams continuously, keeps aggregates current under heavy updates, and serves point queries at up to 30,000 QPS per node, so decisioning flows read live data instead of a batch from the night before.
Can VeloDB handle on-chain data?
Yes. AVE.ai ingests roughly 5,000 records per second from major blockchains with end-to-end latency under 5 seconds, and serves close to 1,000 QPS with P99 under 1 second during peak trading.
What workloads can teams consolidate on VeloDB?
Time-series analysis, OLAP aggregations, OLTP-style point lookups, log search, and vector retrieval run in one engine. BCEL serves roughly 1,000 daily teller and management reports plus ad-hoc analytics from a single cluster after retiring its Hadoop stack; Advance.AI unified log analysis with analytics.
How does VeloDB deploy for regulated financial institutions?
Managed cloud, BYOC in your own cloud account, or on-premises, with storage-compute separation for elastic scaling and full control over data residency.
