4 ms·

The modern way is to sidestep the issue altogether and use Kubernetes with a database designed to run on Kubernetes. You can get sharding, replication and leade

The modern way is to sidestep the issue altogether and use Kubernetes with a database designed to run on Kubernetes. You can get sharding, replication and leader election essentially for free - you can concentrate on using the database instead of running the database. Compute is really cheap compared to engineering man-hours.

cpucycling7 2y ago
> The modern way is to sidestep the issue altogether and use Kubernetes Kubernetes does require quite some time to learn/master. So you could say one replaces one time-consuming issue with another.
remram 2y ago
What's "a database designed to run on Kubernetes"? Cassandra?
- fastest963 2y ago
  Something like Yugabyte or Cockroach
- linuxdude314 2y ago
  Pretty sure they were referring to the operators you can install that will stand up production ready replicated clusters. They work quite well.
  - remram 2y ago
    They replied to a comment mentioning "pgBouncer, repmgr, patroni" so supposedly not one of those.
- meow_catrix 2y ago
  [dead]
- dgraph_advocate 2y ago
  Dgraph is designed to run on Kubernetes.
  - remram 2y ago
    Not relational, not a replacement for Postgres. Also doesn't look open-source at first glance.
    - dgraph_advocate 2y ago
      You should try it before claiming ”not relational”, as it can totally store and use relational data, and also be very useful. Even though it says ”graph” you don’t need to write graph traversal queries - just describe in GraphQL the nested data model you want to get out of the database and that’s what you will get. Community edition is open source.
      - remram 2y ago
        > Community edition is open source. If that is true, you could definitely not make it less obvious. This is not the impression I got from your repo ("variously licensed under the Apache Public License 2.0 (APL) and the Dgraph Community License"). Neither your docs nor your website mensions a "community edition" either.
        dgraph_advocate 2y ago
        It is not my project and I have nothing to do with Dgraph. After 15+ years of RDBMS, I just prefer it. Also, first Google hit is dgraph.io: > The only open source, AI-ready graph database that gives developers the tools to quickly build distributed applications at scale.
dionian 2y ago
it was a minor pain finding and setting up a postgres operator in k8s, but once i got it going it wasn't too horrible. are these other solutions that are more built for it significantly easier to manage?
- slig 2y ago
  Which one did you end up choosing?
kachapopopow 2y ago
https://artifacthub.io/packages/helm/bitnami/postgresql https://artifacthub.io/packages/helm/bitnami/postgresql, postgres is fortunately one of them. Downside is that you cannot scale writes.
yrro 2y ago
Does the Cloud Native PG operator count?
- cheboygan 2y ago
  +1 on checking out cloud native PostgreSQL operator, or other PG operators like crunchy or zalando or ongres

cpucycling7 2y ago

> The modern way is to sidestep the issue altogether and use Kubernetes Kubernetes does require quite some time to learn/master. So you could say one replaces one time-consuming issue with another.

remram 2y ago

What's "a database designed to run on Kubernetes"? Cassandra?

fastest963 2y ago
Something like Yugabyte or Cockroach
linuxdude314 2y ago
Pretty sure they were referring to the operators you can install that will stand up production ready replicated clusters. They work quite well.
- remram 2y ago
  They replied to a comment mentioning "pgBouncer, repmgr, patroni" so supposedly not one of those.
meow_catrix 2y ago
[dead]
dgraph_advocate 2y ago
Dgraph is designed to run on Kubernetes.
- remram 2y ago
  Not relational, not a replacement for Postgres. Also doesn't look open-source at first glance.
  - dgraph_advocate 2y ago
    You should try it before claiming ”not relational”, as it can totally store and use relational data, and also be very useful. Even though it says ”graph” you don’t need to write graph traversal queries - just describe in GraphQL the nested data model you want to get out of the database and that’s what you will get. Community edition is open source.
    - remram 2y ago
      > Community edition is open source. If that is true, you could definitely not make it less obvious. This is not the impression I got from your repo ("variously licensed under the Apache Public License 2.0 (APL) and the Dgraph Community License"). Neither your docs nor your website mensions a "community edition" either.
      - dgraph_advocate 2y ago
        It is not my project and I have nothing to do with Dgraph. After 15+ years of RDBMS, I just prefer it. Also, first Google hit is dgraph.io: > The only open source, AI-ready graph database that gives developers the tools to quickly build distributed applications at scale.

dionian 2y ago

it was a minor pain finding and setting up a postgres operator in k8s, but once i got it going it wasn't too horrible. are these other solutions that are more built for it significantly easier to manage?

slig 2y ago
Which one did you end up choosing?

kachapopopow 2y ago

https://artifacthub.io/packages/helm/bitnami/postgresql https://artifacthub.io/packages/helm/bitnami/postgresql, postgres is fortunately one of them. Downside is that you cannot scale writes.

yrro 2y ago

Does the Cloud Native PG operator count?

cheboygan 2y ago
+1 on checking out cloud native PostgreSQL operator, or other PG operators like crunchy or zalando or ongres