# Design draft: a low I/O piece storage

**URL:** <https://forum.storj.io/t/design-draft-a-low-i-o-piece-storage/24864>\
**Category:** Product Discussions\
**Created:** [January 7, 2024, 7:19pm UTC](https://forum.storj.io/t/design-draft-a-low-i-o-piece-storage/24864 "2024-01-07T19:19:38Z")\
**Posts on this page:** 1\
**Showing post:** 26

<div class="post-metadata">

**Author:** ![s-t-o-r-j-user](https://storj.bcdn.literatehosting.com/user_avatar/forum.storj.io/s-t-o-r-j-user/32/7816_2.png) [@s-t-o-r-j-user](https://forum.storj.io/u/s-t-o-r-j-user)\
**Post date:** [January 20, 2024, 10:07am UTC](https://forum.storj.io/t/design-draft-a-low-i-o-piece-storage/24864/26 "2024-01-20T10:07:59Z")

</div>

I just skimmed twice (in a positive sense) through your paper. Am I correct that you are suggesting storenode to be mostly relying on “new metadata database” instead of on underling filesystem? A kind of architecture similar to `Lustre FS` metadata servers `(MDS)` and metadata targets `(MDT)` but on a micro scale?

---

_[View the full topic](https://forum.storj.io/t/design-draft-a-low-i-o-piece-storage/24864)._