techhub.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
A hub primarily for passionate technologists, but everyone is welcome

Administered by:

Server stats:

4.6K
active users

#distributedsystems

1 post1 participant0 posts today

An easy question about internet message headers (RFC 822 and subsequent revisions).

Take a message thread where each reply has a single parent message.
Such a thread is often shown as a tree, linking replies by their `References' header fields.
Why does _each_ such field contain _all_ parent message ids up to the "root" message of the thread?
Wouldn't the id just of the parent message be enough, as in the `In-Reply-To' header field?

#DistributedSystems
#InternetMessages
#SoftwareEngineering

Die Race Conditions im Wissensmanagement findet man nicht nur bei KI-Nutzern.

"Bei der Analyse ist herausgekommen, dass mehr als jede dritte Seite problematisch sein soll. Mindestens 20 Prozent der Einträge enthielten demnach Informationen, "die nicht mehr aktuell sind". Nur bei der Hälfte davon steche dies Nutzern sofort ins Auge."

heise.de/news/Wikipedia-wegen-

heise online · Wikipedia wegen vieler Fehler und veralteter Informationen in der KritikBy Stefan Krempl
#aigen#noAI#knowledge

🚨 Multi-Master PostgreSQL gotchas: INSERT/INSERT conflicts can silently lose data, UPDATE/UPDATE overwrites entire tuples (not just changed columns), and 3-node clusters risk permanent divergence.

New blog post from Shaun Thomas dives deep into CRDTs (Conflict Resistant Data Types 😉), sequence safety, and conflict resolution strategies 👇

pgedge.com/blog/living-on-the-

Understanding Eventsourcing: Planning and Implementing scalable Systems with Eventmodeling and Eventsourcing leanpub.com/eventmodeling-and- by Martin Dilger is the featured book on the Leanpub homepage! leanpub.com #SoftwareEngineering #EventDrivenArchitecture #SoftwareArchitecture #Software #DistributedSystems #books #ebooks

The first book to combine Eventmodeling & Eventsourcing to plan software systems of any size and complexity.

Find it on Leanpub!

LeanpubUnderstanding EventsourcingThis book gives a detailed introduction to Eventsourcing and how to apply it in real world scenarios with a focus on implementation.
Continued thread

When to Choose Cluster vs. Single-Node? 🤔
The cluster version is specifically designed for scenarios where horizontal scaling across multiple machines is necessary because you're hitting those single-host limits.

Check more information in our latest announcement 👇
hubs.li/Q03sRKrg0
#Observability #Logs #OpenSource #DevOps #DistributedSystems #VictoriaMetrics #monitoring

VictoriaMetricsVictoriaLogs: Scalable | Open Source | Logs DB & Logging SolutionVictoriaLogs is a fast and easy-to-use, open source logs solution. Highly scalable on cloud, kubernetes or on-premise setups.

New work:

Distributed Transaction Patterns

github.com/ha1tch/dxp/blob/mai

A framework for understanding and implementing distributed transaction patterns through a phase-based spectrum approach.

The DXP documentation is organized as a progressive journey through distributed transaction patterns.

Distributed Transaction Patterns. Contribute to ha1tch/dxp development by creating an account on GitHub.
GitHubdxp/README.md at main · ha1tch/dxpDistributed Transaction Patterns. Contribute to ha1tch/dxp development by creating an account on GitHub.

💡 Rethinking ? Listen to the with Sam Newman for key lessons:

🔹 Use Microservices as a last resort, as distributed systems bring serious complexity
🔹 Architecture is a Continuous Journey, not a fixed destination
🔹 Focus on the Outcome
🔹 Build in Observability from day one to enable better decision-making
🔹 Microservices are not binary, but a gradient. It's OK to have microservices for part of your systems, while the rest remain monoliths

🔗 bit.ly/3FYjFyr

Constellation update: the system is at the point of a bare-minimum end-to-end demo. I can submit a basic request to a peer, have it send to a consensus pool (if needed), then to a processor, then report the result back.

The "last mile" of code is janky and has lots of tech-debt to get to this point on a certain timeline, but it's there. The core architecture is mostly sound, though.