TigerBeetle (@TigerBeetleDB) header
TigerBeetle (@TigerBeetleDB) profile picture

The financial transactions database designed to power the next 30 years of transaction processing.

Joined May 2021
Photo by TigerBeetle (@TigerBeetleDB)Photo by TigerBeetle (@TigerBeetleDB)Photo by TigerBeetle (@TigerBeetleDB)Photo by TigerBeetle (@TigerBeetleDB)Photo by TigerBeetle (@TigerBeetleDB)Photo by TigerBeetle (@TigerBeetleDB)Photo by TigerBeetle (@TigerBeetleDB)Photo by TigerBeetle (@TigerBeetleDB)Photo by TigerBeetle (@TigerBeetleDB)Photo by TigerBeetle (@TigerBeetleDB)
TigerBeetle retweeted
Every time I dig deep into DB's and their architectures, @TigerBeetleDB always impresses me. I bow.
1
2
182
3. TigerBeetle’s architecture Read how a database designed for financial transactions approaches contention, replication, durability, and retries. github.com/tigerbeetle/tiger…
1
3
55
2,243
TigerBeetle retweeted
System design prompt: design Stripe payments ledger. Start with requirements: 1) Immutable, auditable history. No UPDATE in place. Every balance change is an event. 2) Correctness over latency. Strong consistency per account. Idempotent writes. 3) Queries: current balance, balance at time T, list transactions, reconciliation/export. 4) Multi-currency, fees, refunds, disputes, chargebacks, holds. APIs + data model: - POST /entries (idempotency_key, account_id, currency, amount, type, external_ref) - POST /transfers (from_account, to_account, amount, currency, ref) - GET /balance?account_id=&as_of= Tables: - accounts - journal_entries(id, ts, currency, metadata, external_ref) - entry_lines(entry_id, account_id, amount, direction, available_vs_pending) Balance is sum(lines) grouped by account+currency. Keep derived balance snapshots for speed, but treat them as cache. Architecture: - Write path: API -> validator -> append-only store (SQL with constraints) -> async projector builds balance snapshots and search indexes. - Partition by account_id so all mutations for an account serialize on one shard. - Use unique(idempotency_key, account_id) to make retries safe. - Every transfer is double-entry: two lines, same entry_id, commit atomically. Scaling + tradeoffs: - Hot accounts: shard within account by sub-ledger (available/pending) or by currency, but keep a single ordering key. - Read performance: periodic snapshots (eg every 1k entries) + replay tail. - Exactly-once is a trap. Build at-least-once with idempotency + reconciliation jobs. Failure cases to discuss: - Partial writes: enforce DB transaction for entry + lines, or nothing. - Duplicate requests: idempotency keys must survive deploys and cache evictions. - Clock skew: ordering is DB sequence, not client timestamps. - Backfills and reversals: never delete. Post compensating entries. - Projector lag: balance endpoint can serve strongly consistent (compute from source) vs fast (snapshot) with clear staleness bounds.
2
7
630
TigerBeetle retweeted
A neat little trick I've been using at work to write cross-platform scripts that run on Windows (PowerShell) and Linux. Stolen from @TigerBeetleDB
2
3
483
Talking less agents, more distributed systems. Having a blast.
Day one of distributed systems therapy (the real DST) with @jorandirkgreef and the @TigerBeetleDB team in Brazil. Consensus, Object Storage, Deterministic Simulation Testing on the Sugarloaf. Muito bom 🇧🇷
1
2
25
1,255
Just use TigerBeetle.
Replying to @0xlelouch_
Solid breakdown. 🔥 One concern from the payment backend side: concurrent withdrawals. Imagine an account has $100. Two $80 withdrawals arrive simultaneously. Both read the same balance before the async projector catches up. Without atomic balance checks and serialization per account, both could succeed. I’d keep balance validation and ledger writes in the same DB transaction, with row-level locking or conditional updates. Async projections? Great for reporting. Not for authorizing withdrawals. How would you handle this without sacrificing throughput?
1
5
57
1,936
TigerBeetle retweeted
Just use TigerBeetle
System design prompt: design Stripe payments ledger. Start with requirements: 1) Immutable, auditable history. No UPDATE in place. Every balance change is an event. 2) Correctness over latency. Strong consistency per account. Idempotent writes. 3) Queries: current balance, balance at time T, list transactions, reconciliation/export. 4) Multi-currency, fees, refunds, disputes, chargebacks, holds. APIs + data model: - POST /entries (idempotency_key, account_id, currency, amount, type, external_ref) - POST /transfers (from_account, to_account, amount, currency, ref) - GET /balance?account_id=&as_of= Tables: - accounts - journal_entries(id, ts, currency, metadata, external_ref) - entry_lines(entry_id, account_id, amount, direction, available_vs_pending) Balance is sum(lines) grouped by account+currency. Keep derived balance snapshots for speed, but treat them as cache. Architecture: - Write path: API -> validator -> append-only store (SQL with constraints) -> async projector builds balance snapshots and search indexes. - Partition by account_id so all mutations for an account serialize on one shard. - Use unique(idempotency_key, account_id) to make retries safe. - Every transfer is double-entry: two lines, same entry_id, commit atomically. Scaling + tradeoffs: - Hot accounts: shard within account by sub-ledger (available/pending) or by currency, but keep a single ordering key. - Read performance: periodic snapshots (eg every 1k entries) + replay tail. - Exactly-once is a trap. Build at-least-once with idempotency + reconciliation jobs. Failure cases to discuss: - Partial writes: enforce DB transaction for entry + lines, or nothing. - Duplicate requests: idempotency keys must survive deploys and cache evictions. - Clock skew: ordering is DB sequence, not client timestamps. - Backfills and reversals: never delete. Post compensating entries. - Projector lag: balance endpoint can serve strongly consistent (compute from source) vs fast (snapshot) with clear staleness bounds.
1
2
47
3,333
Just use TigerBeetle.
System design prompt: design Stripe payments ledger. Start with requirements: 1) Immutable, auditable history. No UPDATE in place. Every balance change is an event. 2) Correctness over latency. Strong consistency per account. Idempotent writes. 3) Queries: current balance, balance at time T, list transactions, reconciliation/export. 4) Multi-currency, fees, refunds, disputes, chargebacks, holds. APIs + data model: - POST /entries (idempotency_key, account_id, currency, amount, type, external_ref) - POST /transfers (from_account, to_account, amount, currency, ref) - GET /balance?account_id=&as_of= Tables: - accounts - journal_entries(id, ts, currency, metadata, external_ref) - entry_lines(entry_id, account_id, amount, direction, available_vs_pending) Balance is sum(lines) grouped by account+currency. Keep derived balance snapshots for speed, but treat them as cache. Architecture: - Write path: API -> validator -> append-only store (SQL with constraints) -> async projector builds balance snapshots and search indexes. - Partition by account_id so all mutations for an account serialize on one shard. - Use unique(idempotency_key, account_id) to make retries safe. - Every transfer is double-entry: two lines, same entry_id, commit atomically. Scaling + tradeoffs: - Hot accounts: shard within account by sub-ledger (available/pending) or by currency, but keep a single ordering key. - Read performance: periodic snapshots (eg every 1k entries) + replay tail. - Exactly-once is a trap. Build at-least-once with idempotency + reconciliation jobs. Failure cases to discuss: - Partial writes: enforce DB transaction for entry + lines, or nothing. - Duplicate requests: idempotency keys must survive deploys and cache evictions. - Clock skew: ordering is DB sequence, not client timestamps. - Backfills and reversals: never delete. Post compensating entries. - Projector lag: balance endpoint can serve strongly consistent (compute from source) vs fast (snapshot) with clear staleness bounds.
4
11
175
7,761
TigerBeetle retweeted
I've built a block commit protocol for a JSON parser for fuzzing via DST. Tiger style is exponentially valuable. json-scanner-playground.fly.…
1
4
609
We are actively rewriting software w/ TigerStyle & DST at Tempo. This is the role I am the most excited to hire right now. Please reach out. Need someone extremely AI pilled who also gets that "correct software" is the prerequisite to hillclimbing for performance.
looking for a DST guru to join the Tempo team - lmk if anyone's on the market and keen to write high assurance software that can move billions of dollars in milliseconds securely
22
10
5
322
28,987
Real programmers allocate once.
36
41
11
671
59,287
TigerBeetle retweeted
The way @TigerBeetleDB documented all on youtube i loved it :) building components, discussion around databases there so much more to learn form it youtube.com/@tigerbeetledb/p…
1
3
569
Day one of distributed systems therapy (the real DST) with @jorandirkgreef and the @TigerBeetleDB team in Brazil. Consensus, Object Storage, Deterministic Simulation Testing on the Sugarloaf. Muito bom 🇧🇷
6
5
1
97
5,149
If you were building a database from scratch today instead of 30 years ago (when most of the popular ones were built), you could pick any advancements in architecture, hardware, language, or research to implement. Where would you start? amplifypartners.com/blog-pos…
2
7
1,091
TigerBeetle retweeted
Definitely @TigerBeetleDB . It’s a no brainer.
2
3
684
going to make all my out there experiments in Zig TigerStyle Zero Deps w/ DST & report back - first project is a replacement to aerospace (the window manager), codename tatami
11
3
2
81
11,317