HackerTrans
TopNewTrendsCommentsPastAskShowJobs

vmarchaud

no profile record

Submissions

Why did we choose to self manage Kubernetes in 2022 for a small startup?

vmarchaud.substack.com
1 points·by vmarchaud·4 anni fa·0 comments

comments

vmarchaud
·2 anni fa·discuss
Jumping the opportunity:; we make a product that do hyper-personalisation for marketing emails, we do have few customers using SFMC and we tried to think about some use cases with Agentforce but i don't see a lot of useful things. Some customers want to see data about who saw personalisation in tableau/datacloud but again this isn't direct integration with Agentforce but more sending data to any salesforce product. Did you see any pratical use case for Agentforce in SFMC ? Thanks
vmarchaud
·3 anni fa·discuss
If someone else doesn't understand why painters would have a problem with Tesla (like me), you might be interested in this: https://www.ifmetall.se/aktuellt/tesla/background-informatio...
vmarchaud
·3 anni fa·discuss
They do have a policy to only include free software [1]:

> We only accept formulae that use a Debian Free Software Guidelines license or are released into the public domain following DFSG Guidelines on Public Domain software into homebrew/core.

[1]: https://docs.brew.sh/License-Guidelines
vmarchaud
·3 anni fa·discuss
I've written about this specific subject here [0], feel free to reach out or ask them here if you have other questions

[0]: https://vmarchaud.fr/deploying-kubernetes-the-hard-way/
vmarchaud
·3 anni fa·discuss
You could do that with Ansible, there's kubespray [0] from the official project to do that. You may want to read [1] starting with [2]. There's also k3s [0] and k0s [4]. I've written about my setup here [5] also if you are interested.

[0]: https://github.com/kubernetes-sigs/kubespray

[1]: https://www.tauceti.blog/

[2]: https://www.tauceti.blog/posts/kubernetes-the-not-so-hard-wa...

[3]: https://k3s.io/

[4]: https://k0sproject.io/

[5]: https://vmarchaud.fr/deploying-kubernetes-the-hard-way/
vmarchaud
·3 anni fa·discuss
You can follow [0] which is currently stage 2 to fix this

[0]: https://github.com/tc39/proposal-async-context
vmarchaud
·3 anni fa·discuss
I've used OpenTelemetry since it's original alpha in 2020. Originally the main issue i had was supporting tracing across common libraries (there wasn't a lot of libraries supported back then). Now (and i recently worked with it) i would say is which protocol is supported by which component: your sdk generate spans/metrics in a specific format, then you send that to a collector that accept a range of protocol versions and finally you can send that to your vendor ... but you need to know which protocol/version it supports.

That's not actually something you can do much about considering the sheer size of opentelemetry (both in term of implementation and vendors working on it) and i expect for people implementing nowadays, proto should be pretty stable and my experience should theorically not be the case anymore.
vmarchaud
·3 anni fa·discuss
Someone from F5 worked on this with OpenTelemetry [0] for Arrow, another effort was done for Parquet but was dropped [1]

[0]: https://github.com/open-telemetry/oteps/pull/171

[1]: https://github.com/open-telemetry/opentelemetry-proto/pull/3...
vmarchaud
·3 anni fa·discuss
We have ~two dozens rented boxes @ OVH in their French datacenters, apart from hardware failures from time to time (1/2 per year) you should be fine, just take that into account while designing the architecture. Also don't expect the support to helps you, they will sometimes but i'll say it's better to not expect it to not be surprised :)
vmarchaud
·3 anni fa·discuss
You might be interested into Matt Mochary free books [0], he's a succesfull ceo coach around the valley.

[0]: https://docs.google.com/document/d/1ZJZbv4J6FZ8Dnb0JuMhJxTnw...
vmarchaud
·3 anni fa·discuss
Hey, i'm interested in the kind of problems you have run against in those cases if you don't mind sharing them ? My email is in my bio
vmarchaud
·3 anni fa·discuss
Not specially something that you cannot do without WASM but at $WORK we are using a WASM build of Skia [0] to render canvas from nodejs.

Why use WASM ? Because we wanted to stay close to our stack (ie. calling wasm from nodejs). Do it work ? Yes, memory consumption is quite heavy though (each WASM module have its own heap that can quickly grow).

However we are looking to directly use Skia now and avoid the overhead of WASM so i think its a nice solution in the beginning but you might want to ditch it later on.

[0]: https://github.com/google/skia/tree/main/modules/canvaskit