Highlights & Quotes: DevOps, Naming Things, Mimi Makes A Million, Message Queues (Disque)

Post date: Sep 5, 2015 5:41:59 PM

DevOps book:

You can’t have DevOps and still have separate operations and development teams. Period.

Some days you’re writing code, other days you’re doing a deployment, or maintenance. Everyone shares all of those responsibilities equally.

debugging a production issue.

“non deveoper” skillset, be that design, project managment or whatever.

continuous integration and zero-downtime

DevOps is indeed bullsh*t. What matters is good people working together without artificial boundaries. Tech is tech.

Naming things:

The web of names, hashes and UUIDs 12 Mar 2015

DHT (Distributed Hash Table) such as Kademlia or Chord and let it work its magic.

UUID

Antirez weblog Adventures in message queues

message broker and it’s called Disque.

Redis masters (without replication or clustering whatsoever)

Redis HA / Cluster

Travelling Rodos:

Traffic lights will be installed till the end of 2013 which will mark the completion of the road making left turns available.

A jersey barrier in the middle.

Mimi Makes A Million, no quotes just read.

Message Queues (Disque):

By default Disque runs in-memory only, and uses synchronous replication to achieve durability (however you can ask, per message, to use asynchronous replication).

Is queueing partially/totally consistent across a group of servers or divided up for maximal throughput?

however message ordering is preserved in a best-effort way.

Can consumers and producers look into the queue, or is it totally opaque?

Is queueing unordered, FIFO or prioritised? Best-effort FIFO-ish as explained.

Named queues. Producers and consumers does not need to coordinate, since nodes use federation to discover routes inside the cluster and pass messages as they are needed by consumers.

Is message posting transactional? Yes,