Editorials, Encryption/Data Security

CLR Capability in SQL

SSWUGtv
With Stephen Wynkoop
In todays issue of SSWUGtv Steve interviews Dan Clark as he shares his thoughts bout self-service BI. Is PowerPivt the tool you should be using? Find the answers here on this edition of SSWUGtv. Watch the Show

CLR Capability in SQL
We have been discussing the value of the CLR in SQL Server for the last couple of editorials. Today we have a response to my thought that the CLR can be an asset for distributed processing taking advantage of the sharding technique built into SQL Azure federations.

Veska Writes:
I both agree and disagree with your comments on CLR.

  • I do not think it is the job of code to navigate the database for the information it needs. SQL can do this beautifully. Most commercial programming consists of screens + database inputs and outputs if you keep the two separate, i.e. the screens do not need to know the structure of the database and the database just receives the information collected by the screens and returns any required information doing all the necessary navigation of tables. This gives you full freedom to design, tune and keep the database as you need it for either commercial of performance reasons.
  • I have not found anything that you cannot code in SQL , except for someone deciding to store information in a blob, that only a piece of code can decode.



Thanks Veska. I think you pretty well clarify why I am ambivalent to the use of CLR capabilities in SQL Server. There are many situations where if TSQL is adequate, then use it. Even when TSQL is not adequate, I am not always convinced that the database engine is the place to put procedural code. One key factor for me is determining what places the least load on the database engine. If it is less work to do something in the database than to return data to be processed elsewhere, then maybe you should consider doing the work in the database engine.

Lately, as I have been reviewing Big Data techniques, I am disappointed that the engines surrounding BIG Data always assume it must be un-structured, and push stuff into storage that may not be as optimized as a relational database. Try and find a map reduce capability built into a sharded relational database structure. I’m thinking that is what SQL Azure federations could be if it included CLR support.

Who knows? Perhaps that is just too academic. Well, tomorrow we move on to new territory with patterns in JavaScript. Drop me a note with your comments at btaylor@sswug.org.

Cheers,

Ben

$$SWYNK$$

Featured White Paper(s)
Encryption & Key Management for Microsoft SQL Server 2008
Written by Townsend Security

Simplify encryption and key management on … (read more)

Featured Script
Create_Add_Proc
This procedure will look to the systems tables and create a default add procedure for a table… (read more)