SQL Server and SAS

[VDD]

Good day or evening,

I have a question that would be related to the future growth in our company for using SAS rather than SQL server.

For 19 years I have been working in the same position for many directors and managers who have came and gone.

SAS was a requirement when I started and has always been until the last 2 years. 

It seems that the newbie's coming into the department are pushing SQL server as the DB to use. While in my department they are in the process of now converted 100's of SAS programs into SQL, maybe because the new users just don't understand the building blocks for creating, cleaning and manipulating data.

My question: is SAS losing out the MS SQL server?

[RW9]

I wouldn't say they are losing out as such.  SAS is not a database, its a programming language with built in stats analysis modules.  It has a different structure and method of processing to a database.  There isn't any exclusivity, you could use various databases and alongside that have SAS.  In fact, most companies should be looking at some sort of datawarehousing/etl type tool to hold all data, whilst simultaneously aligning reporting or analysis tools to take data from said warehouse, that is the logical path.

As for changing SAS code into SQL, unfortunately that is a trend most of us have seen.  Even SAS are pushing it more and more with various systems.  Personally I don't see why, there doesn't seem to be that much to gain.  Sure SQL is used in lots of other systems so is very useful for connections and some processing, but other than some powerful joining there doesn't seem to be a huge benefit.

SAS is however losing out somewhat to other analysis tools like R, and in the sense of data scientist type scenarios to Python.  However that is always the case, and there is no reason again why those ecosystems cannot flourish around each other.

[ballardw]

---

@VDD wrote:

Good day or evening,

I have a question that would be related to the future growth in our company for using SAS rather than SQL server.

For 19 years I have been working in the same position for many directors and managers who have came and gone.

SAS was a requirement when I started and has always been until the last 2 years. 

It seems that the newbie's coming into the department are pushing SQL server as the DB to use. While in my department they are in the process of now converted 100's of SAS programs into SQL, maybe because the new users just don't understand the building blocks for creating, cleaning and manipulating data.

My question: is SAS losing out the MS SQL server?

---

Obviously we don't know every task you might use SAS for.

I might suggest asking one of the SQL Server proponents how to do a regression analysis or complex sample design survey analysis with SQL if you do any of those. Or almost anything statistical more complex than Proc freq or simple Proc means summaries (max min mean std). I suspect even basic correlations in raw SQL would be entertaining.

And while on the topic of SQL server ask them if they are storing date values in datetime type variables. I seem to see an awful lot of "date" values that all have a time of 00:00:00 appended. Which I always think means that somewhere there is a lot of wasted disk space. I may quite well be wrong on that bit but looking a dozens of columns half of each populated with 00:00:00 is tedious.

[RW9]

Par exemplar:

https://communities.sas.com/t5/SAS-Programming/Summary-Statistics-for-Different-Variables-for-differ...

[VDD]

a lot of datetime with the 0's and a lot of varchar date junk.  While as a SAS programmer we can always clean the data which is something that is required in all of our SAS task that need to reference those DB's.  Thanks for the insight.

[SASKiwi]

The company I work for is pretty much a "Microsoft Shop" and in fact SQL Server is the prime source of data for our SAS applications. Fortunately though there is also a strong appreciation of what SAS brings to the table. The type of processing and calculations we do in SAS would be either extremely difficult to do in SQL or pretty much impossible. So SQL Server and SAS peacefully co-exist and each is used for what they are best at for the most part.

I use SQL in SAS a lot primarily to build datasets that then need the row-by-row processing and conditional logic of the SAS DATA step. You can't do that in SQL Server unless you start building Stored Procedures containing cursors. Not only do these often perform really badly but they are near impossible to maintain except by the original developer because  the coding is so complicated compared to SAS.