# Deploy cloud native full stack operator without CSI driver

**URL:** <https://community.dynatrace.com/t/deploy-cloud-native-full-stack-operator-without-csi-driver/48075>\
**Category:** Container platforms\
**Tags:** dynatrace-operator, kubernetes\
**Created:** [October 1, 2024, 2:59pm UTC](https://community.dynatrace.com/t/deploy-cloud-native-full-stack-operator-without-csi-driver/48075 "2024-10-01T14:59:43Z")\
**Posts on this page:** 1\
**Showing post:** 3

<div class="post-metadata">

**Author:** ![Dant3](https://dub1.discourse-cdn.com/dynatrace/user_avatar/community.dynatrace.com/dant3/32/159_2.png) [@Dant3](https://community.dynatrace.com/u/Dant3)\
**Post date:** [October 1, 2024, 6:34pm UTC](https://community.dynatrace.com/t/deploy-cloud-native-full-stack-operator-without-csi-driver/48075/3 "2024-10-01T18:34:56Z")

</div>

Actually, you can do it without a CSI. Is called application only. And then a CRD with Host monitoring only.

But it would create an storage issue, since each pod would have an unique lib of the modules, as stated here:

> **[How it works](https://docs.dynatrace.com/docs/ingest-from/setup-on-k8s/how-it-works)**
>
> In-depth description on how the deployment on Kubernetes works.

Each pod would download the libs. If each lib goes around 100MB with 10 pods you would have 1GB used in the storage only for libs., additionally, unless you have a local copy (own repo) of the libs, each pod would take its time to download the required lib from the cluster.

So is not that you can’t… is just no efficient compared with a CSI pod per node and all using that share volume.

---

_[View the full topic](https://community.dynatrace.com/t/deploy-cloud-native-full-stack-operator-without-csi-driver/48075)._
