SQL and which stack works best

I was asked by a colleague a few days ago, how much SQL do you need to know to working production high demand environment? I said, you need at least 5 years advanced, 10 years intermediate, and overall 15 years of SQL experience. She was surprised.

 

SQL in the LAMP stack is one thing. SQL in the Microsoft stack is another. SQL is SQL. On both, it depends on your platform, in and out, processing, and overall performance requirements. For me, OLE-DB and native SQL is the best. 8x to 50x performance boost over layering, translation, and for me, I like direct calls, even when fetches are really just better done in sequences, batches, and other things. Just don't lock up a db and table on a cursor read. 

 

On premise? in a client/server situation? LAN/WAN? JSON REST XML AJAX? Or in the cloud on AWS or AZURE? It doesn't really matter. It's the performance constraints that matter. The architecture matters. Indexing matters. Security and protection at all layers in the stack. Encryption and redundancy, hot, warm, cold storage, live data, streaming data, archival, post processing, batching, ETL situations; other integrations; alarms, monitoring, backups, restores, snapshots, you name it, stored procedures. It all matters. There are maxims to each of these and best practices, and roadmaps, and runbooks (we used to call it ETL procedures), from bcp to install, fully unattended documented procedures that another could follow and it would just work.

 

I explained, one of my mentees, some 5 years ago, I trained in 4-6 months what took me 15 years to master. He picked up the torch and burned the midnight oil, and voila! He was only 25 years old and for me, I was 45 years old. He went onto be a senior developer starting from nothing, just really fast to pickup on things. Now he is working in GO programming at NASA and GE, and prior to that, I helped him hone is C# skills and SQL in a high frequency trading platform. 6 months of master sensei and I smiled at him and gave him the nod.

 

If we can mentor so many quickly then we all win. I don't care if they surpass me. I surpassed my mentor(s) after 15 years, and so will many others do the same to me. I love to help, share, learn, and provide great results that everyone enjoys and lasts as long as necessary - stands the test of time.

 

For me, I have used Microsoft SQL Server the longest, since 1993, and MongoDB since 2017, PostgreSQL since 2002, MariaDB since 2008, and mySQL since 2014. Its an odd grouping of everything. From 1GB to 1TB datasets. From 1TB to 8TB of offload and its all amazing it stays hinged today and no CRC errors.

 

Like I always said, I love to crush data.  PostgreSQL i like more and more. MSSQL still love it.

 

more to come...