Launch Week Blog Post: Billion event querying

[andyvan-ph]

Issue to dicuss and coordinate blog post for the upcoming launch week: see issue.

• Copy Deadline: March 16
• Publish Date: March 22
• Outline: Deep dive into how we developed the platform to handle massive event loads (tentative)

To Do

• Blog post to be written by @macobo with support from @paolodamico and @andyvan-ph
• Blog artwork from @lottiecoxon

[andyvan-ph]

Note: outline is only placeholder, so feel free to suggest more interesting approaches.

[macobo]

The original plan was to follow https://github.com/PostHog/product-internal/pull/240 and document that.

Of that issue, we will have managed to build out at the time of the release:

• 1. Improved deployments (helm chart refactoring done by @guidoiaquinti and I)
• 2. Upgrading clickhouse
• 3. Sharded clickhouse on self-hosted
• 6. Schema: Including event in the events sort key

In addition we've:

• Had a few painful experience scaling out postgres and debugging it. cc @guidoiaquinti for that side of things
• Made it possible to run posthog with Altinity Cloud
• Introduced a new async migrations tooling for larger data migrations going forward (cc @yakkomajuri / @tiina303)
• Internally constructed a plan for changing product semantics and how person data is stored Future of our person model meta#39
• person_distinct_id2 for everyone

Do you think that's enough material for the blog post? What to focus on in there?

Sadly we haven't gotten to the biggest bits due to a lack of resources/time/prerequisites.

cc @marcushyett-ph

[marcushyett-ph]

Given we still have a way to go to hit the seamless Billion event querying milestone... I would advocate for focussing this article on:

How we improved query performance by X% for our largest projects

We can split this into a few short technical sections, describing the motivation and the nuances of our solutions, i.e:

• new person_distinct_id table
• Schema
• Sharded clickhouse

I think it's worth having a separate post on deployments and helm chart refactoring @andyvan-ph @guidoiaquinti (e.g. "How we simplified our 10 service cross-platform self-hosted deployment")

[andyvan-ph]

RE: above, 'How we improved query performance by X% for our largest projects' sounds good to me if everyone's happy with that?

I think it's worth having a separate post on deployments and helm chart refactoring @andyvan-ph @guidoiaquinti (e.g. "How we simplified our 10 service cross-platform self-hosted deployment")

Yup, agreed. That's covered in #3067

[lottiecoxon]

idea for this I wanted to run past you @andyvan-ph, for the new title of query performance - I thought it would be fun to do a sport scientist scene with hedgehogs, one running on a treadmill and the others in lab coats assessing how to make improvements?

[andyvan-ph]

idea for this I wanted to run past you @andyvan-ph, for the new title of query performance - I thought it would be fun to do a sport scientist scene with hedgehogs, one running on a treadmill and the others in lab coats assessing how to make improvements?

Sounds great, Lottie. Go for it

[lottiecoxon]

Done! on Figma ready for export when you need them

[andyvan-ph]

@lottiecoxon thanking you 👍

[paolodamico]

Deadline for this is in two days, @macobo need any help getting this out the door?

[macobo]

Nothing written yet focussing on projects.. Was planning to draft this tomorrow and Wednesday.

[guidoiaquinti]

I’ll be back tomorrow from vacations. i should have a draft as well soon

…

On Mon, 14 Mar 2022 at 12:37, Karl-Aksel Puulmann ***@***.***> wrote: Nothing written yet focussing on projects.. Was planning to draft this tomorrow and Wednesday. — Reply to this email directly, view it on GitHub <#3066 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA6Z3GKP5PREXRDX2PBS7VLU74QGBANCNFSM5PULULQA> . You are receiving this because you were mentioned.Message ID: ***@***.***>