Recap: OSDI London meetup, 20 March

On 20th March we held the latest London OSDI meetup in CentrEd at ExCel, just round the corner from, and just after the end of, the Kafka Summit conference.

The meeting was co-hosted by Aiven and decodable.

The session was organised as lightning talks, short talks on any topic that would be interesting to the attendees, although naturally we suggested something related to Apache Kafka®

We had eight talks in all, three of them volunteered at the meeting itself. @ftisiot ordered them randomly for us (it seemed the fairest solution):

  • Robin Moffat from decodable introduced us to fzf, the fuzzy finder, which he finds incredibly useful at the command line - check it out.

  • @tibs_at_aiven (that’s me) explained how to send Avro messages using a
    scheme registry (see demo6 at Fish & Chips & Apache Kafka), and also enthused about the fact one can now create topics using the aiokafka library.

  • AJ Danelz from aklivity introduced us to the Zilla project, and in particular the mqtt-simulator which makes it easy to simulate the production of MQTT messages by IoT sensors (:heart: the message specification).

  • Gunnar Morling from decodable talked us through his kcctl command line tool, which makes it much simpler to manage Apache Kafka® Connect. It’s styled after kubectl. I was particularly taken by the use of auto-completion. A unique feature is that it can patch connector offsets, normally regarded as rather difficult :frowning:

  • Aaditya Talwai of Lariat Data talked us through “Circuit Breakers and Graceful Degradation using Flink Streaming Primitives”. He made a compelling case.

  • @BenGamble7 talked about building a phone game using webrtc and Kafka,
    for use at a conference, with unreliable WiFi and very few resources.

  • Jove Zhong of timeplus talked about Timeplus Proton, their approach to combining streaming and historical storage (with the power of Kafka and ClickHouse).

  • Konstantinos (Dean) Dalianis told us about nile which can autogenerate Kafka clients for binary formats from human-understandable descriptions in YAML files, using Kaitai Struct. For instance, the LoRa BE-ICS-CE
    RisingHF
    format (who doesn’t :heart: well-designed auto-code generation).

I loved all of these, and was so happy at the variety of talks offered.

3 Likes