# If google would use dgraph, should it use for everything one single dgraph DB, or for every service (Maps, YouTube, GMail...) an own dgraph DB?

**URL:** <https://discuss.dgraph.io/t/if-google-would-use-dgraph-should-it-use-for-everything-one-single-dgraph-db-or-for-every-service-maps-youtube-gmail-an-own-dgraph-db/15824>\
**Category:** Dgraph\
**Tags:** kind:question, dgraph\
**Created:** [October 17, 2021, 9:08pm UTC](https://discuss.dgraph.io/t/if-google-would-use-dgraph-should-it-use-for-everything-one-single-dgraph-db-or-for-every-service-maps-youtube-gmail-an-own-dgraph-db/15824 "2021-10-17T21:08:59Z")\
**Posts on this page:** 1\
**Showing post:** 6

<div class="post-metadata">

**Author:** ![Juri](https://yyz1.discourse-cdn.com/flex007/user_avatar/discuss.dgraph.io/juri/32/8872_2.png) [@Juri](https://discuss.dgraph.io/u/Juri)\
**Post date:** [October 19, 2021, 1:26am UTC](https://discuss.dgraph.io/t/if-google-would-use-dgraph-should-it-use-for-everything-one-single-dgraph-db-or-for-every-service-maps-youtube-gmail-an-own-dgraph-db/15824/6 "2021-10-19T01:26:34Z")

</div>

thanks a lot!!!

> [@amaster507](#):
>
> HA still allows you to link data/edges across clusters just not across tenants.

the problem is as our buddies said:

> [@MichelDiz](#):
>
> Dgraph will balance the predicates based on disk usage.

> [@Feature Request: Making Learner Nodes more smart/useful with stopping wasting resources/money. + We need smart learning-sharding like Cloudflare Durable Objects](https://discuss.hypermode.com/t/feature-request-making-learner-nodes-more-smart-useful-with-stopping-wasting-resources-money-we-need-smart-learning-sharding-like-cloudflare-durable-objects/15681/2):
>
> The automatic rebalancing system in dgraph is based on disk usage per predicate, and I feel like a better algorithm could be used to distribute based on usage instead of disk size, but that may be a different issue all together.

If I do everything into one database, that will result on scale into a terrible performance. or not? because dgraph doesn’t know that I e.g: want social media stuff into one alpha node and ecommerce stuff into the other alpha node. dgraph will just balance them randomly based on the disk usage. that means that everything will be mixed up. then my joins have to cross abroad to multiple alpha nodes(=machines), that will create a terrible latency. or not?

it would be really awesome to have different big projects in one single world wide dgraph database, that would allow awesome query capabilities. but if dgraph won’t shard stuff logically(recognize which data is used frequently together to shard between services like socialmedia and ecommerce), but instead based on disk usage which will result in a mix that leads to terrible performance latency and so on. or not?

---

_[View the full topic](https://discuss.dgraph.io/t/if-google-would-use-dgraph-should-it-use-for-everything-one-single-dgraph-db-or-for-every-service-maps-youtube-gmail-an-own-dgraph-db/15824)._