While we're all recovering from the outage / heart-attack. FYI that I have a support ticket about TERRIBLE performance with CSV imports since our OCI migration (PHX). Our CSV import times more than doubled since OCI. We do routine imports of 5k, 10k records (pricing updates, etc) and the throughput is about 40-45% of what it used to be (and it wasn't "fast" before OCI). They are actually hearing me out and have been investigating it for several weeks now. As well as requesting more before/after sample CSV timings. If anyone else is noticing the slower CSV import performance since OCI, I would HIGHLY encourage you to file a ticket. Be ready to provide basic info (number of records, duration before and after, etc). I think this is tied to overall slow performance with OCI and the need for them to improve performance.

Going to get us a ticket on this, horrible upload speeds. Sometimes our 13 uploads of 7k items will take a week!!

just curious, are you running scripts/workflows on the import or not? or does that seem to have a big impact if you've tried both?

We are not running scripts/workflows on the import. I had checked that some weeks ago, on a tip from someone here.

Does the performance fluctuate depending on the time of day? May be worth testing that as it could be also something to do with the load on the shared infrastructure.

Late reply. Unfortunately running the CSV imports at odd-hours of the morning/night does not seem to change the throughput speed very much. It seems to only churn through 100-125 (max) record per minute. It feels more like intentional throttling than a performance bottleneck. NS is still "investigating" and collecting information.

Mmm, what I'd look at next is see if there is a direct correlation to running the update via a Map\Reduce script (1 and 2 processors). You would need to do this in your sandbox if you have one. Start with checking CSV import speeds in the sandbox to get a baseline. As you say, it sounds like throttling on the CSV import functionality, so your own Map\Reduce 'might' bypass it.

@Dan Curtis +1 here Since OCI, our imports and performance seems to have dipped. Also lots of “50x Server not available errors” in our data integrations. It would make total sense for Oracle to go economical on “compute’ power when migrating to sell more SuiteCloud+ licenses.