it turns out congestion pricing is actually quite good for many businesses because saving hours of wasted labor and gas stuck in traffic is far more valuable than the $9 charge
Watch Nadella describe SaaS apps as nothing more than a CRUD database with some business logic, but once the business logic moves to AI agents, SaaS is over:
@SQLGene Not sure this is related in a direct sense, though VertiPaq came out around this time so maybe?
Either way writing with V-order is proprietary and available in Fabric’s writer engines, but openly readable.
@JohannesVink@Ko_Ver Agreed. The time consuming part of Power BI ingestion is VertiPaq compression, not reading from storage. This will help but not be a game changer.
@bob_duffy@PawarBI I’ll be curious to see how the access control around running a python runtime notebook works with SPNs and managed identities, along with any auto recommendations around shifting notebooks from a Spark runtime to the python runtimes.
@bob_duffy@PawarBI I’ve seen a looootttt of cases where a spark cluster is just running orchestration code or glue code that never touches Spark, and while it’s absolutely necessary for ELT. And wastes plenty of VM just because a managed spark cluster was the only way to run some scheduled python.
@bob_duffy Though there’s some options with Azure Relay, but haven’t used that in a while. When you register a gateway with a fabric tenant it needs outbound traffic for that initial connection, over documented ports for what turns out to be azure relay.
@bob_duffy For proxies I’ve got nothing as there are so many ways that could be set up. Been using virtual network appliances (VNAs) lately given client environments that handle the NAT portion for traffic to the “internet”.
@JohannesVink@databricks It’s extremely annoying! In azure the blob inventory report can be used to get this for you, and that can be a parquet file that you make into a delta table, then query it in UC, but also why on earth isn’t this all handled for you.