Indexes and costs

Misc
1

I am a fan of anyone that is willing to produce a pay per drink database service with good performance that scales. It lowers the barrier to small players to do big things. If I had to pay for every GB of index that needs to sit in memory for the life of my product, well thats not the kind of innovation I am looking for basically.

Slash is pay per GB transferred and scales with transfer vs overall size of indexes.This is currently the main factor after graph orientation that motivates me to use Slash GraphQL over other services. I would love to see more focus on this, maybe a blog post and also assurances we will not get slapped with a change in model down the road would be really nice.

Can you clarify what is unique about Dgraph that enables you to make this pricing available?
Can you illustrate the performance of Dgraph and especially Slash as index sizes scale?
Does Slash face any cold-start issues as perhaps workers spin up or data is paged into memory?

Finally along the lines of reducing runtime cost for operating large indexes, I noticed this gem today on Hacker News PGM Indexes: Learned indexes that match B-tree performance with 83x less space | Hacker News perhaps this is applicable to Dgraph indexes?

Thanks in advance for any insights!

1 Like
2

Note that there’s a limit on how much one backend can store on Slash GraphQL (shared instances) before they are required to upgrade to Dgraph Cloud (dedicated instances). @gja can confirm the limit per backend.

Slash would be using multi-tenancy, so we can fit many backends on the same bigger database — hence amortizing the cost of operations for us, in a way that’s sustainable for us business wise.

1 Like
3

@mrjn thanks for the clarification on that. Any documentation on where that break happens? Some more in-depth coverage of managed services performance and tco at various scales would be a great addition to the existing docs / blog posts. THANKS!

1 Like
4

Hi There,

We are currently allowing users to load upto 25gb of data into their Slash GraphQL instance. Beyond that, you may upgrade to our dedicated instances.

Tejas

1 Like
5

This topic was automatically closed 41 days after the last reply. New replies are no longer allowed.