# Scalability of deployments

**URL:** https://hub.mender.io/t/scalability-of-deployments/257
**Category:** General Discussions
**Tags:** yocto
**Created:** [February 23, 2019, 1:30am UTC](https://hub.mender.io/t/scalability-of-deployments/257 "2019-02-23T01:30:43Z")
**Posts on this page:** 4
**Page:** 1

<div class="post-metadata">

### Author: ![matzewagner](https://avatars.discourse-cdn.com/v4/letter/m/958977/32.png) [@matzewagner](https://hub.mender.io/u/matzewagner)
#### Post date: [February 23, 2019, 1:30am UTC](https://hub.mender.io/t/scalability-of-deployments/257/1 "2019-02-23T01:30:44Z")

</div>

We are looking into using mender for OTA updates for our portable backup devices. We want the update process to be generally user-driven and are considering the option of creating a new deployment for every device (for every update). Will creating this many deployments (as opposed to batching devices into fewer deployments) pose a significant overhead that would lead to scalability issues?

---

<div class="post-metadata">

### Author: ![erikhh](https://yyz2.discourse-cdn.com/flex036/user_avatar/hub.mender.io/erikhh/32/102_2.png) [@erikhh](https://hub.mender.io/u/erikhh)
#### Post date: [February 26, 2019, 11:11am UTC](https://hub.mender.io/t/scalability-of-deployments/257/2 "2019-02-26T11:11:43Z")

</div>

I think the main concern would be that the MongoDB database fills up.  
But since MongoDB is a good solid and well maintained database. You could easily write a simple batch job to clear out old deployment records every once in a while.  
Or you could upgrade to a sharded MongoDB setup I guess.

---

<div class="post-metadata">

### Author: ![mirzak](https://yyz2.discourse-cdn.com/flex036/user_avatar/hub.mender.io/mirzak/32/2056_2.png) [@mirzak](https://hub.mender.io/u/mirzak)
#### Post date: [February 26, 2019, 3:47pm UTC](https://hub.mender.io/t/scalability-of-deployments/257/3 "2019-02-26T15:47:26Z")

</div>

I would also image that you lose out a bit in the the GUI, since the deployments are not in batches and they will pile up fast and hard to trace/manage.

The actual download should not be a problem and is the same as it would have been in a “batched” deployment.

---

<div class="post-metadata">

### Author: ![eystein](https://avatars.discourse-cdn.com/v4/letter/e/e5b9ba/32.png) [@eystein](https://hub.mender.io/u/eystein)
#### Post date: [February 28, 2019, 10:07pm UTC](https://hub.mender.io/t/scalability-of-deployments/257/4 "2019-02-28T22:07:24Z")

</div>

In general this should not be a problem, I am aware of users with thousands of devices that create individual deployments for each device (automated, of course).

Also, we have an upcoming feature in the mid-term “dynamic deployments” which should resolve issues like these. In that case you’d just create one deployment and devices will join/leave that deployment based on dynamic conditions (like users confirming, etc.).
