Thanks for another great video! Apart from the base image choice which I see it's being addressed, what I feel I miss the most is the lack of GitOps support. I'm ok with having DBAs managing their own configs and settings the way they're used to but not interacting directly with clusters. Apart from that I agree with pros and cons you already listed.
Hey Viktor, great video once again! This one is actually right on time for me, since I want to start implementing HA for a bunch of databases that I manage and already run as single replicas in k8s. As it's only three different types of databases, mariadb, postgres and mongodb namely, I would have gone with a custom solution for each one. Maybe you just saved me a lot of work on that, so thank you :) I'm saying maybe as I am too a bit worried about the Ubuntu-thing, but if there's at least a hint that the project might move to a more sane base image in the future, I'll give it a shot... Happy holidays! 🎄
Most base images have been moved to alpine in KubeBlocks 0.8.1 and KubeBlocks-Addons, some base images are still Ubuntu as they are built by other third-party database providers, we are also pushing them to adopt the more secure solutions. Thanks for your suggestions.@@TheLolle97
To be fair with ubuntu, the latest tag (updated 18 days ago) scanned with trivy gets this CVE score Total: 13 (UNKNOWN: 0, LOW: 12, MEDIUM: 1, HIGH: 0, CRITICAL: 0). So we're talking about 13 issues not 300 and they are low to medium without available fixes. Also the image size is 28MB. That being said, I'm 100% with you when you say that images should only have the bare minimum needed to run the app so I agree that Alpine is a better choice anyway
I think that CNPG is more powerful, partly because it was designed specifically for PostgreSQL (in Kubernetes). Where KubeBlocks shines is not necessarily as a competition to PostgreSQL but, rather, as a way to run different types of databases in a uniform way (and in Kubernetes).
@@DevOpsToolkit Ok that makes sense. If we're only doing PG probably makes more sense to go with specific operator. But if we're doing different databases and want a standard way of hosting them, this would probably be better.
KubeBlocks allows you to run all sorts of databases in Kubernetes. "Big data" expression, on the other hand, tends to indicate that there is a lot of data. Data needs to reside in a database (or file system).
Having a ton of CRDs is cumbersome and annoying to many users. If it had 80 CRDs that would be the first reason I would ditch it. Otherwise great video !
The development with kubernetes gets more and more messy. It's like a medicine with some byeffects for that you need more and more medicine to compensate the byeffects in a chain. End of the the will end up in a CRD hell 😀
What do you think about KubeBlocks?
I wish they'd support mssql as well 😢
@pavelpikat8950 mssql is closed source so i don't think they would be allowed to support it.
It's funny that I got this recommended to me, because I'm right now exactly at the point where I want to add a DB to my personal cluster.
Great overview with pros and cons as always. Thank you and happy holidays!
Thanks for introducing KubeBlocks!
Thanks for another great video! Apart from the base image choice which I see it's being addressed, what I feel I miss the most is the lack of GitOps support. I'm ok with having DBAs managing their own configs and settings the way they're used to but not interacting directly with clusters. Apart from that I agree with pros and cons you already listed.
Hey Viktor, great video once again! This one is actually right on time for me, since I want to start implementing HA for a bunch of databases that I manage and already run as single replicas in k8s. As it's only three different types of databases, mariadb, postgres and mongodb namely, I would have gone with a custom solution for each one. Maybe you just saved me a lot of work on that, so thank you :)
I'm saying maybe as I am too a bit worried about the Ubuntu-thing, but if there's at least a hint that the project might move to a more sane base image in the future, I'll give it a shot...
Happy holidays! 🎄
I suspect that KubeBlock people already saw the video and will take an action regarding the base image.
Let's hope they do 😄 @@DevOpsToolkit
I am an engineer from KubeBlocks, we are working on the base image replacements from Ubuntu to Alpine, and thank you for you suggestions.
Most base images have been moved to alpine in KubeBlocks 0.8.1 and KubeBlocks-Addons, some base images are still Ubuntu as they are built by other third-party database providers, we are also pushing them to adopt the more secure solutions. Thanks for your suggestions.@@TheLolle97
To be fair with ubuntu, the latest tag (updated 18 days ago) scanned with trivy gets this CVE score Total: 13 (UNKNOWN: 0, LOW: 12, MEDIUM: 1, HIGH: 0, CRITICAL: 0). So we're talking about 13 issues not 300 and they are low to medium without available fixes.
Also the image size is 28MB.
That being said, I'm 100% with you when you say that images should only have the bare minimum needed to run the app so I agree that Alpine is a better choice anyway
How does this compare to database specific operators like CloudnativePG? And why would I use this over that?
I think that CNPG is more powerful, partly because it was designed specifically for PostgreSQL (in Kubernetes). Where KubeBlocks shines is not necessarily as a competition to PostgreSQL but, rather, as a way to run different types of databases in a uniform way (and in Kubernetes).
@@DevOpsToolkit Ok that makes sense. If we're only doing PG probably makes more sense to go with specific operator. But if we're doing different databases and want a standard way of hosting them, this would probably be better.
Thank you - Do Kubeblocks and big data serve the same purpose?" OR Can Kubeblocks be used for big data processing?
KubeBlocks can be used to manage a database that stores data (big or small).
cmpg for postgres db? can you please clarify I can not search anything related
It's this one: cloudnative-pg.io/
Do Kubeblocks and big data serve the same purpose?
KubeBlocks allows you to run all sorts of databases in Kubernetes. "Big data" expression, on the other hand, tends to indicate that there is a lot of data. Data needs to reside in a database (or file system).
# TIL
## NADA
### CVE
Having a ton of CRDs is cumbersome and annoying to many users. If it had 80 CRDs that would be the first reason I would ditch it. Otherwise great video !
The development with kubernetes gets more and more messy.
It's like a medicine with some byeffects for that you need more and more medicine to compensate the byeffects in a chain.
End of the the will end up in a CRD hell 😀
What would be the alternative to CRDs? How would we run a specialized workload?
Oh I see you are a person that complains that he has nothing to complain about something perfect! 🤪😬