diff --git a/get-started/get-help.md b/get-started/get-help.md
index f241acfa7..445d531f8 100644
--- a/get-started/get-help.md
+++ b/get-started/get-help.md
@@ -600,7 +600,41 @@ Do not use a `XS_` prefix.
 
 ## MTXS
 
-### I get a 401 error when logging in to MTXS through App Router { #mtxs-sidecar-approuter-401}
+### My MTX sidecar is killed with 'Exit status 137'
+
+In this case, the process was killed by a `SIGKILL` signal, typically because it exceeded its resource limits, for example memory or CPU, causing the container platform to terminate it.
+
+::: tip Distinguish extensibility and non-extensibility scenarios
+While out-of-memory issues are more common, with **extensibility enabled** you’re more likely to run into CPU bottlenecks due to expensive compilations that need to be performed at (MTX) runtime.
+:::
+
+MTX uses **four parallel workers** by default to perform tenant upgrades. If your project exceeds a certain complexity threshold, you might run into these resource bottlenecks. We advise you to **follow this algorithm** to mitigate resource overload:
+
+1. **Decrease your model complexity**: Ask yourself, is your current domain model a good compression of your business domain? Decreasing complexity here will have positive trickle-down effects, including tenant upgrade performance.
+2. **Increase resources (scale up)**: Increase the RAM assigned to your MTX sidecar or upgrade task. This is typically done in deployment resources like _mta.yaml_ (Cloud Foundry) or _values.yaml_ (Kyma).
+
+   [Learn more about database upgrade task configuration](../guides/multitenancy/#update-database-schema){.learn-more}
+
+   ::: info In Cloud Foundry, CPU shares scale with memory
+   As there is no way to increase CPU independently from memory, your memory configuration might be a bottleneck even if the process is killed due to CPU spikes.
+   :::
+3. **Decrease workers in async MTX operations**: When scaling up resources is no longer feasible, you can run with fewer parallel migrations:
+    ```jsonc
+    "cds": {
+      "requires": {
+        "multitenancy": {
+          "jobs": {
+            "workerSize": 3 // default: 4
+          }
+        }
+      }
+    }
+    ```
+    > This won't affect application runtime performance.
+
+4. **Increase the number of MTX sidecars (scale out)**: To compensate for eventual performance losses from **3.**, distribute the work across multiple sidecars.
+
+### I get a 401 error for `cds login` through App Router { #mtxs-sidecar-approuter-401}
 
 See [How to configure your App Router](../guides/extensibility/customization#app-router) to verify your setup.
 Also check the [documentation about `cds login`](../guides/extensibility/customization#cds-login).
diff --git a/guides/multitenancy/index.md b/guides/multitenancy/index.md
index 45a780b77..0520d855e 100644
--- a/guides/multitenancy/index.md
+++ b/guides/multitenancy/index.md
@@ -705,6 +705,14 @@ cf run-task <app> --name "upgrade-all" --command "cds-mtx upgrade '*'"
 ```
 :::
 
+::: info Managing large upgrade workloads
+
+Very large projects might need to increase resources or limit parallelism of tenant upgrades.
+
+[The best practice algorithm is laid out in our _Get Help_ guide.](../../get-started/get-help#my-mtx-sidecar-is-killed-with-exit-status-137){.learn-more}{style="margin-top:10px"}
+
+:::
+
 
 ### Test-Drive with Hybrid Setup