Want to make your Python app resilient to any failure?. For the past two months, I've been developing DBOS Python to try to make it as lightweight and easy as possible to add fault tolerance to your program. I'm really happy we've just released and I can share what we've built

RT @petereliaskraft: From talking to engineering teams, I think one of the toughest parts of operating an AI agent is observability. Debugg….

From talking to engineering teams, I think one of the toughest parts of operating an AI agent is observability. Debugging a conventional program is hard enough, but AI agents are even more difficult to reason about because they take actions based on unpredictable, sometimes

RT @qianl_cs: Let's Go! 📣 Join us next week for the July DBOS User Group Meetup. We're unveiling an early preview for DBOS Go, one of our….

Case study:

When you're training thousands of students daily on complex enterprise software, you need reliable infrastructure. Ontologize, who provides hands-on training for clients including PwC, Conagra, and the US Air Force, faced a major challenge: seamlessly and automatically deploying.

RT @qianl_cs: New Screaming in the Cloud episode! Corey Quinn and our CEO @jedberg "chat about Jeremy's "build for three" rule, a plan fo….

Demo:

Just put together a fun new demo: an AI agent that researches Hacker News’s opinion of a topic. The agent starts with a topic, finds relevant stories and comments, analyzes them, then chooses related topics to explore until it has enough data. Then, it synthesizes everything

RT @P99CONF: Most systems rely on external orchestration because it's well-established. But it introduces overhead, write amplification, an….

RT @petereliaskraft: The most elegant thing about durable workflows? Observability comes for free. Durable workflows work by representing….

RT @qianl_cs: Happy weekend! I upgraded and "modernized" our cloud subscription backend to ESM + Fastify + DBOS TypeScript V3 (preview). Fe….

The most elegant thing about durable workflows? Observability comes for free. Durable workflows work by representing your program as database rows. There’s one row for each workflow and step, tracking its status, timing, and outcome. This info is stored for reliability, but

Blog post:

I often get asked how hard it is to add durable execution to an application. Many developers get the value of durable execution. The idea is to checkpoint the state of an application to a persistent store (like a database) so that if the application ever fails it can use its.

Cool new feature: queue deduplication!. When you enqueue a task on a durable queue, you can set a deduplication ID. DBOS guarantees only one workflow with that ID runs at once. Where is this useful? Let’s say your workflows manage customer data, and you only want one workflow to

RT @petereliaskraft: Testing distributed systems at scale is brutal. There's a near-infinite space of potential problems arising from diff….

RT @qianl_cs: The South Bay Systems meetup is back on July 22!. We're excited to welcome @jacopotagliabue (co-founder and CTO of Bauplan)….

Blog post:

Testing distributed systems at scale is brutal. There's a near-infinite space of potential problems arising from different failures happening across different processes at different timings. Unit and integration tests can’t cover this–they struggle to test more than the happy.

RT @qianl_cs: CXL technology is promising for memory disaggregation in databases. This paper explores CXL for memory disaggregation in data….