A good week ago I hosted the monthly T-SQL Tuesday blog party. I invited the community the blog about the idea of what would happen if we would need to go back on-premises, after a good decade of cloud computing. What would you need to do? How do you plan for this? What are the skills you might need to (re-)learn? You can find my take on this here, where I’m telling my junior colleagues that troubleshooting issues on-prem is more challenging, because it’s possible you need to go way down, all the way to Kerberos, SAN and DNS issues.
And the community had some great insights as well! In no particular order:
- Andy got my subtle reference to Back to the Future, and he talks about the capabilities you might need for an on-prem solution. He shares that he’s reassured by the fact that on-prem is still on-prem, and that answers can be found from StackOverflow to LLMs and everything in between. Personally, he feels comfortable with a hybrid solution, it’s not black or white.
- Mike pulls some deep magic blog trick out of his hat: the top 10 list.
- Deborah muses about the fact that not everyone has been in the cloud, that database engine work remains largely the same and that people should finally learn to tune queries
- Brent shares some insights as he actually had clients that moved back on-prem after their cloud journey. The one thing that hit a nerve with me is that hardware has become ridiculously expensive, thanks to AI. (Not only big servers are affected by this, but personal laptops as well)
- Steve coins the term “un-migrating”. He believes some people will love coming back on-prem, but the one thing that will certainly be more difficult is HA.
- Alexander compares the cloud and on-prem as assembling and building, using Lego. There’s always room for Lego. He also thinks about what skills he would need to relearn, and it mostly likely will be troubleshooting.
- Reitse thinks many would struggle to migrate back on-prem, and his advice is to start planning. What do you need, who do you need and what technical knowledge would be required?
- Sander talks about the hidden costs of on-prem, and gives us an approach on how to tackle a possible conversation about migrating. He also loves the cloud because it’s easy to use with Infrastructure-as-Code, and on-prem you might have servers that are the same, but actually not the same.
- Rob is usually the first to blog on T-SQL Tuesday as he’s from Australia (although some cheated by posting before the 9th). He thinks about how sovereignty concerts might impact companies in both Europe and Austrialia, and if new people need to learn something is probably disk-related stuff such as corruption, back-ups and disaster recovery. He also wonders how we would need to deal with all those SaaS providers.
- Kevin had quite a unique take: the Fabric multi-verse. What if Fabric had to be built using separate services? What if Fabric could be bought as an appliance (remember SQL Server Parallel Data Warehouse)? Or if it was voice-controlled?
- Rich lists the things he would worry about, as a server/database performance tuner. Disks, HA/DR, maintenance and virtualization. And also vendors. He states that the main driver will ultimately will be money.
- Chad reflects back on his days in the server room and manages to sneak in a Back to the Future reference. Well done.
Thanks to all the contributors, and I hope to see you all next month for the 200 anniversary edition!
The post T-SQL Tuesday #199: Roundup first appeared on Under the kover of business intelligence.