I noticed that the Serverless pricing calculator has recently changed and shows the number of Namespaces as a multiplier for the base cost.
If I remember correctly, the previous calculator indicated that with more Namespaces, the monthly costs tend to decrease (instead of increasing). This could be a crucial difference for my business where each tenant has a dedicated namespace with varied usage.
It may be just the confusing wording (equating “Namespaces” to “Workloads”), but I’d like to ask for a clarification: how does the number of Namespaces affect the pricing for serverless indexes?
In terms of pricing, we have not changed anything.
We did find that people were getting confused with namespaces on the pricing calculator, so the change was to hopefully clear that up.
The other change was the increase in the default value for the metadata size so that will also have effect.
Now, you would describe the use case at the namespace level, and then the total number of namespaces you plan to have. We recommend using namespace when building multitenant applications on Pinecone.
It may be just the confusing wording (equating “Namespaces” to “Workloads”), but I’d like to ask for a clarification: how does the number of Namespaces affect the pricing for serverless indexes?
To expand on this further, if you had 1 million vectors in 1 namespace, then querying would be more expensive than if you had 10 namespaces with 100,000 vectors each. The storage cost and upsert should be the same.
That all said, we have temporarily removed the calculator; there will be a new, improved one coming soon.
Now, you would describe the use case at the namespace level, and then the total number of namespaces you plan to have
This is the confusing part (at least in my business) - because each tenant has different usage, it’s not easy to come up with one, “average” namespace for calculations.
Meanwhile, I understand that with more namespaces the pricing “stabilizes” and depends more on I/Os so I won’t worry about it for now
