May 2, 2018 at 11:46 am
jasona.work - Wednesday, May 2, 2018 5:48 AMMy mindset has always been, use the tools that are right for the job that your team can support. So, if SQL2017 Linux could be shown to be the right tool for the job, and the team had the knowledge in place to support Linux, then yes.
My view as well. I'd use whatever.
May 2, 2018 at 11:48 am
I use SQL Server on Linux. For supported features, everything has just worked and I still connect to and work on SSoL from a Windows machine. As Solomon mentioned, you don't necessarily change the desktop tools for developers or even DBAs to Linux.
May 2, 2018 at 12:34 pm
If you're like me and work in a larger sized IT organization where the on-prem server hardware and VMs are provisioned and managed by a dedicated Platform Operations / DevOps team, then from the perspective of a DBA, SQL Server is just a managed service. Occasionally I RDP into the host to install SSIS or copy some event logs, but it's not the core of what I do and could be offloaded to DevOps anyhow. For those who are using Azure SQL, Microsoft could have already switched your host VM from Windows Server Core over to Linux and you don't even know it.
"Do not seek to follow in the footsteps of the wise. Instead, seek what they sought." - Matsuo Basho
May 2, 2018 at 12:36 pm
I honestly think Microsoft would rather be out of the OS business altogether. That isn't where the money is at. Besides, with the focus on Azue (which is ran on Linux) why wouldn't they eventually follow Apple and port to an linux kernel? When you are apps chasing market share and going after Oracle, it makes sense.
May 2, 2018 at 1:10 pm
Forget about Windows and Linus, it's theoretically possible that SQLOS could be retrofitted to run alone on bare metal.
"Do not seek to follow in the footsteps of the wise. Instead, seek what they sought." - Matsuo Basho
May 2, 2018 at 8:25 pm
Eric M Russell - Wednesday, May 2, 2018 12:34 PMFor those who are using Azure SQL, Microsoft could have already switched your host VM from Windows Server Core over to Linux and you don't even know it.
Pretty much where I sit on this. I'm in the cloud and it really doesn't matter to me anymore.
I do however still use Linux. Applications and ETL are written in Python on Linux with Azure SQL, Azure Data Warehouse, or Azure Data Lake. This collaboration of both Azure (Microsoft) and open source (Python, Linux) is pretty powerful.
May 2, 2018 at 11:48 pm
xsevensinzx - Wednesday, May 2, 2018 8:25 PMApplications and ETL are written in Python on Linux
Is your Python ETL home grown or are there 3rd party frameworks and tools involved? I'm looking at ETL tools per se and asking myself what is the real benefit. It looks like Python and Airflow could do much of what I need.
Ive written a PySpark based ETL framework and am happy that it does the majority of what I actually need. It took a bit of work but now allows new sources to be added quite quickly
May 3, 2018 at 6:10 am
David.Poole - Wednesday, May 2, 2018 11:48 PMxsevensinzx - Wednesday, May 2, 2018 8:25 PMApplications and ETL are written in Python on LinuxIs your Python ETL home grown or are there 3rd party frameworks and tools involved? I'm looking at ETL tools per se and asking myself what is the real benefit. It looks like Python and Airflow could do much of what I need.
Ive written a PySpark based ETL framework and am happy that it does the majority of what I actually need. It took a bit of work but now allows new sources to be added quite quickly
I basically built what Airflow is from scratch before I found out what it was. Then I saw it really had no support for Azure anyways, just AWS. Not that it matters as you can just build your own Azure connectors with it. I would surely take a good solid look at Airflow, it looks nice.
The reason I went this route because most of my data is third-party (advertising data) and always requires some type of data connector with an API. In meaning, I can do all the development of said data connector in Python, which is cheaper and easier to do versus doing it in .NET. It's also easier to do distributing ETL workloads in Python across more than one node for a single job with no additional costs other than whatever the VM costs. The other thing too, is I work in data science as the tech resource, so being able to also integrate the ML piece was nice. Ends up being a 3-and-1 approach all within the same language that can be easier to version control and everything.
What you mentioned about PySpark is nice too. I use the same with Azure with Azure Data Lake Analytics (U-SQL). This ends up being another level of ETL that you can rely on with the flow. Python to do some, U-SQL to do some in the lake, then finally doing some in the targeted system like SQL Server/Data Warehouse. You can really distribute the workload across all 3 platforms, which really helps you balance the load.
May 3, 2018 at 7:12 am
After migrating some of our databases to Azure, I installed SQL Server 2016 Enterprise Edition and SSIS on an Azure IaaS host and consolidated all of our ETL packages there. The BI team also has SSAS and some 3rd party tools running there as well.
But, one of these days I'll have to get into Python and see what all the hype is about.
"Do not seek to follow in the footsteps of the wise. Instead, seek what they sought." - Matsuo Basho
May 3, 2018 at 1:10 pm
I can more easily see SQL Server running on Linux in the cloud vs on premise. The main driver is the cost and staff you will need. Linux system administrators usually make a premium over Windows admins. Also more people have skills with Windows Server than with *nix. I look at it a bit like running Java against a SQL Server dataset instead of C# and vise versa.
May 4, 2018 at 10:06 am
Jeff Mlakar - Thursday, May 3, 2018 1:10 PMI can more easily see SQL Server running on Linux in the cloud vs on premise. The main driver is the cost and staff you will need. Linux system administrators usually make a premium over Windows admins. Also more people have skills with Windows Server than with *nix. I look at it a bit like running Java against a SQL Server dataset instead of C# and vise versa.
Depends on your environment. I've worked a few where there were more Linux admins. In that case, maybe this makes more sense.
May 4, 2018 at 10:27 am
Steve Jones - SSC Editor - Friday, May 4, 2018 10:06 AMJeff Mlakar - Thursday, May 3, 2018 1:10 PMI can more easily see SQL Server running on Linux in the cloud vs on premise. The main driver is the cost and staff you will need. Linux system administrators usually make a premium over Windows admins. Also more people have skills with Windows Server than with *nix. I look at it a bit like running Java against a SQL Server dataset instead of C# and vise versa.Depends on your environment. I've worked a few where there were more Linux admins. In that case, maybe this makes more sense.
True. Also, I think it would be wise for everyone (regardless of their thoughts / feelings / experiences, if any, regarding Linux and this new option of SQL Server running on Linux) to keep in mind that this is new territory for this community. For many years there was a heavy anti-Linux sentiment, then a warming up to it but still a deeply ingrained belief that it was unthinkable that SQL Server ever could / would run on Linux. I think it will take time to get the current SQL Server community (a large enough percentage of it) to do a complete 180 and welcome it. I think it's asking a lot of many people to go from disparaging something to embracing it, just because someone up the food chain changed directions. Most people don't come around that quickly, if ever. Just look at how long it took for various civil rights movements to progress through the US (and elsewhere), and how long some current struggles are taking. The option for SQL Server on Linux will likely be more acceptable to non-Microsoft folks, but even there the sentiment has been quite anti-MS for a long time. It might take years for this to gain decent momentum, but I believe that it certainly will, especially with the younger folks who haven't had 20+ years of Linux or MS bashing to condition their thoughts and perceptions.
Take care,
Solomon...
SQL# — https://SQLsharp.com/ ( SQLCLR library ofover 340 Functions and Procedures)
Sql Quantum Lift — https://SqlQuantumLift.com/ ( company )
Sql Quantum Leap — https://SqlQuantumLeap.com/ ( blog )
Info sites — Collations • Module Signing • SQLCLR
May 4, 2018 at 12:58 pm
Solomon Rutzky - Friday, May 4, 2018 10:27 AMSteve Jones - SSC Editor - Friday, May 4, 2018 10:06 AMJeff Mlakar - Thursday, May 3, 2018 1:10 PMI can more easily see SQL Server running on Linux in the cloud vs on premise. The main driver is the cost and staff you will need. Linux system administrators usually make a premium over Windows admins. Also more people have skills with Windows Server than with *nix. I look at it a bit like running Java against a SQL Server dataset instead of C# and vise versa.Depends on your environment. I've worked a few where there were more Linux admins. In that case, maybe this makes more sense.
True. Also, I think it would be wise for everyone (regardless of their thoughts / feelings / experiences, if any, regarding Linux and this new option of SQL Server running on Linux) to keep in mind that this is new territory for this community. For many years there was a heavy anti-Linux sentiment, then a warming up to it but still a deeply ingrained belief that it was unthinkable that SQL Server ever could / would run on Linux. I think it will take time to get the current SQL Server community (a large enough percentage of it) to do a complete 180 and welcome it. I think it's asking a lot of many people to go from disparaging something to embracing it, just because someone up the food chain changed directions. Most people don't come around that quickly, if ever. Just look at how long it took for various civil rights movements to progress through the US (and elsewhere), and how long some current struggles are taking. The option for SQL Server on Linux will likely be more acceptable to non-Microsoft folks, but even there the sentiment has been quite anti-MS for a long time. It might take years for this to gain decent momentum, but I believe that it certainly will, especially with the younger folks who haven't had 20+ years of Linux or MS bashing to condition their thoughts and perceptions.
Take care,
Solomon...
Increasing levels of abstraction and frameworks has been central to the evolution of IT technology over the past 30 years. This encourages polygot solutions than span multiple languages, database platforms, and vendors.
"Do not seek to follow in the footsteps of the wise. Instead, seek what they sought." - Matsuo Basho
May 5, 2018 at 2:44 am
Solomon Rutzky - Friday, May 4, 2018 10:27 AMTrue. Also, I think it would be wise for everyone (regardless of their thoughts / feelings / experiences, if any, regarding Linux and this new option of SQL Server running on Linux) to keep in mind that this is new territory for this community. For many years there was a heavy anti-Linux sentiment, then a warming up to it but still a deeply ingrained belief that it was unthinkable that SQL Server ever could / would run on Linux. I think it will take time to get the current SQL Server community (a large enough percentage of it) to do a complete 180 and welcome it. I think it's asking a lot of many people to go from disparaging something to embracing it, just because someone up the food chain changed directions. Most people don't come around that quickly, if ever. Just look at how long it took for various civil rights movements to progress through the US (and elsewhere), and how long some current struggles are taking. The option for SQL Server on Linux will likely be more acceptable to non-Microsoft folks, but even there the sentiment has been quite anti-MS for a long time. It might take years for this to gain decent momentum, but I believe that it certainly will, especially with the younger folks who haven't had 20+ years of Linux or MS bashing to condition their thoughts and perceptions.Take care,
Solomon...
Agree++
To me Linux feels a more natural environment for scripted installation and deployment and with it consistency and reliability. As a data person this matters more than pointless religious wars.
Strangely I'm finding MacOS a bit of a pain. Something that works at fine in Linux might not work on a Mac. I'm puzzled that some open-source apps don't work either properly or at all on Macs.
I do prefer the Windows 10UI
Viewing 14 posts - 16 through 28 (of 28 total)
You must be logged in to reply to this topic. Login to reply