Closed samuelstroschein closed 2 months ago
We can go multiple routes:
Big question: What are the numbers for the three options?
My take
Intuitively, I would go for SMB to unblock monetization and larger scale usage but don't get slowed down today by designing for enterprise.
@martin.lysk1 and I agreed on:
Let's lower the bar to 25k variants @martin.lysk🦜 ?
We defined the bar to be 250k variants to avoid building with rxdb which we would need to throw away too soon. With sqlite the scaling limits are lix, a problem we need to fix anyways. For the next 6 months 125mb/10 = 12,5MB (250k variants / 10 = 25k variants) is enough with our current user base. We can think about sclaing lix then AFTER we have proven that lix is generalizable.
_From https://discord.com/channels/897438559458430986/1262694367429267499/1262772334058668143_
Agreed with @martin.lysk1 to lower the requirements to 25k variants.
EDIT: Agreed on
Context
We should agree what scalability requirements that we use to design inlang for over the next 12 months.
Proposal
Agree on number of messages we use in performance testing.
cc @martin.lysk1 @jldec
Additional information