A few questions from a Dgraph Noob

My team is considering using dgrpah to store web data…we are making a decent sized niche search engine. Ok, its not very small…we are definitely looking at billions of pages so what I am interested in is…

How is it decided what gets served on disc vs. ram?

What would be looking at in terms of insert speed? SQL failed big time here. We have tons of data being written and updated.

Is there any example apps that are public that are using the full-text search?

How does the full-text search stack up against ElasticSearch?

Our app requires the storage of stats that are used to create other summary stats that are not time series per se. In both cases, these are both just edge nodes that can be added at any time…correct?

Any and all advice with this is most appreciated.

Thanks!

mkl

1 Like

Dgraph is a persistence-first database, so everything definitely goes to disk. But, we have an LRU cache which can keep a portion of the underlying key-values in memory. You can set how much memory you want to dedicate to the LRU cache by using --lru_mb flag.

I’d say very fast. But, you know, try it out.

We haven’t compared. But, the full-text search is complete. No one has complained about it.

If I understand this right, what you’re asking is, can you add nodes at any time? Yes.

1 Like

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.