Barbarin David

@FranckPachot @PatHelland @SQLServer @PostgreSQL @OracleDatabase Yes but I had in mind optimized locking that improves writer <-> writer part learn.microsoft.com/en-us/sql/rela… . But most efficient MVCC implementation comes with in-memory tables

@mikedavem @PatHelland @SQLServer @PostgreSQL @OracleDatabase Yes, I guess it has improved for Accelerated Recovery

.@PatHelland, from how @SQLServer RCSI works you generalize & claim "Today's Databases Don't Scale"🥶 cidrdb.org/cidr2024/paper… But... no 👉@PostgreSQL doesn't "climb back in history" - it's the opposite 👉@OracleDatabase latch on XCUR to build CR clone is short and stays in cache

#sqlhelp When using #Azure AD authentication + PIM with SSMS, it is boring to restart it to refresh permissions and get access to menus granted by the escalated privilege. Any tips for that ? cc @erinstellato

Topic is not only for DBs by the way :)

Interesting stuff and discussion today ... provisioning an #azure SQL DB with #IaC is easy but protecting against deletion is not (regarding the context) if you want to stay "agile" :)

Finally, we got an answer from Microsoft that database copy to Az Sql behind PE work with AAD authentication and not SQL one.

#sqlhelp I'm using CREATE DATABASE COPY AS OF with SQL Azure DBs behind private endpoints and according to my tests it works well. I didn't enable public access. The BOL (tinyurl.com/y3su9z82) seems to indicate the opposite. Any thoughts, feedback on this topic ?

@browncspence @arpit_bhayani Mongodb only supports async replication. Easy to scale in this case but consistency is not guaranteed across nodes …

@arpit_bhayani Wrong. The most popular NoSQL database, MongoDB, is ACID compliant and scales “beautifully” horizontally.

Run away from system design instructors who say, "Use a NoSQL database because SQL doesn't scale" 🤦‍♂️ It shows the sheer immaturity and lack of practical experience of the instructor because crude generalization doesn't exist in computer science and software engineering. Always remember, ⚡ You get some, you lose some. Relational databases give you strong guarantees and hence they need data to be present in a single node. Non-relational databases do not offer ACID and hence they can easily scale horizontally. So, if you drop constraints and ACID requirements from SQL, you are effectively left with a data shard that you can easily scale horizontally with any routing technique and cluster management. Facebook runs the world's largest MySQL cluster and they have beautifully scaled it. ⚡ I keep writing and sharing these engineering nuggets, so if you are keen on learning them, follow along. 160+ in-depth system design videos youtube.com/c/ArpitBhayani #AsliEngineering #SystemDesign

@CFDevelop @Cameron10300 Performance is another topic but not the argument for splitting a monolith into microservices. By the way, big tables in our context is not a problem. At the opposite, one microservices db with few records (~100) is a very hot spot and good indexing strategy helps to scale

@CFDevelop @Cameron10300 The only performance / scalability argument is probably not relevant here and we miss probably the real context. I work in a microservices environment with sqlserver dbs and main goal is to scale dev teams and reduce the time to market.

A while back, I worked with an SQL server database with several tables that had over a trillion records each, and it ran fine. Please, tell me again why it's necessary to split your database with a few hundred thousand records into #microservice dbs

@erinstellato Application intent=readonly ..

@erinstellato Having an option for connecting to AG RO replica instead of putting the parameter in advanced section

SSMS users, this question's for you... What are the connection options you use most that would be helpful to have on the main page (so not just server, auth, UN & PW)? Too many for a poll, so please reply.. Thank you!!

@FranckPachot Who is jocking here ? your kid or you ? :D

Me to kid, on WhatsApp, talking about his oral exam and my fingers typed "oracle" exam 😭😭😭

@redgate sqlcompare and automation (with powershell) is very 👍. It saves lot of time to check if we miss some objects during database migration especially when there lot of databases …

Finally challenge met : automated provisioning #Azure Sql VM and workgroup AG with terragrunt / #terraform and powershell DSC (through extension) . A good day ….

Moving to the cloud has some advantages and gives priority to old requests and fix some tech debts that would cost in the cloud ^^

#sqlhelp Does anyone knows how to get cause of warning with Azure Data Migration service ? I didn't find out any evidence of information so far . The data transfer is ok but I get a warning warning status at the end ...

@jdanton ... and probably meets most of customer scenarios

@jdanton Unfortunately ARM template seems to not address all the scenarios as far as I know (multi-subnet, workgroup AGs ...) but that's a good start for sure

#Azure SQL IaaS vs PaaS makes really sense when it comes IaC ... Not the same effort for provisioning services... Expected but deserves to be reminded

@jdanton Amount of code (terraform in my case) and settings to deal with for the same "kind of architecture" . SQL availability group is a good example (Azure SQL DB vs SQL MI vs Azure SQL VM)

@mikedavem In what sense?