Craig here from Citus. Happy for any others that have had experience with Citus to chime in, though happy to try to answer them to give some insights while full

Craig here from Citus. Happy for any others that have had experience with Citus to chime in, though happy to try to answer them to give some insights while fully admitting it might be a bit partial :) 1. We've got plenty of customers that rely on Citus as their system of record. How good is it really comes down to your use-case. We're a good fit for certain transactional workloads that have a natural shard key (this is the case for many multi-tenant apps). Or hybrid tranasction/analytical apps like Heap. In these cases you're usually powering some end user facing dashboard and need subsecond query response across a large dataset much like Heap. We're less of a fit for traditional data warehousing, and then if you only have 10 GB of data single node Postgres works just fine. 2. We have customers that come on board for both Enterprise and Cloud. In the past year since we've launched our database-as-a-service we do have a steeper curve of people ramping up on it. It's of course, easier to press a button and have a scalable database than it is to set it up and manage yourself. In general if you're already running on AWS then Citus Cloud makes sense, if you can't run on AWS we're always happy to try to make Enterprise work for you, not to mention there is the open source version. 4. This really varies based on application, lots of apps don't run into issues until they're at a 1-2 TB, and others benefit from parallelism as early as 40-50 GB. That's not to say they couldn't keep throwing more hardware at it, but we've seen some customers migrate at 100 GB and get 2-3x or more performance improvement from single node Postgres. As the rest are a bit more specific to Heap I'll let them chime in, though happy to answer any more questions from the Citus side where helpful.