Gabi Lehner has good news for us:
Azure Data explorer team is constantly focused on reducing COGS and making sure users are paying only for value they are getting.
As part of this initiative, we’re now adding a new automatic capability to stop unused clusters.
In case, you created a cluster and did not ingest any data to it or even if you ingested data and later, you’re not running any queries or ingesting new data for days, we will automatically stop that cluster.
I have two thoughts on this. First, good. Frankly, every cloud service should have automatic pausing unless it makes sense not to—that is, pausing should be the default, not a feature you add later. This is especially true for expensive data processing services.
Second, based on the description, I think I’d like a little more control over this, in terms of how long we go before auto-stop kicks off. It’s ten days, which is a reasonable + large number, but other numbers could make just as much sense for a given user. I like the idea that we see in Databricks and in Azure Synapse Analytics Spark pools: give me a reasonable default, but let me change it in case the reasonable default can’t cut it for some reason.