Simons SQL Blog : SQL Server 2005, Katmai

Whats in a BOL link?

simonsabin — Thu, 16 Aug 2007 21:44:33 GMT

In using BOLyou will have noticed that the links in BOL contain some lovely guids. I often want to reference BOL in documents, forum posts etc but was concerned that the guids might change. So I asked Alan Brewer whether they change or not. He gave me a great response which I've got his permission to publish here.

"There are two issues with GUID reliability, both should be solid.

First, when I asked a year ago, the MSDN publishing teams said they plan to keep supporting both GUID-based and asset ID-based URL forms.

Second, in our authoring system, once a topic is created it retains the same GUID across releases. This is independent of where in any TOC the topic is surfaced. So all SQL Server 2005 topics that still exist in SQL Server 2008 have the same GUID in both environments. We started this policy when we forked the SQL Server 2008 BOL source from the SQL Server 2005 BOL, so does not apply back to the SQL Server 2000 BOL. We plan to carry it forward to future releases.

MSDN/TechNet support many different formats of URLs, including URLs with version identifiers. These URLs take you to the SQL 2005 and SQL 2008 versions of a topic shared between the two BOLs, even though the topic is at different locations in the two different TOC's:

SQL Server 2005

http://msdn2.microsoft.com/en-us/library/ms189084.aspx

http://msdn2.microsoft.com/en-us/library/ms189084(SQL.90).aspx

http://msdn2.microsoft.com/en-us/library/82d7819c-b801-4309-a849-baa63083e83f.aspx

http://msdn2.microsoft.com/en-us/library/82d7819c-b801-4309-a849-baa63083e83f(SQL.90).aspx

SQL Server 2008

http://msdn2.microsoft.com/en-us/library/ms189084(SQL.100).aspx

http://msdn2.microsoft.com/en-us/library/82d7819c-b801-4309-a849-baa63083e83f(SQL.100).aspx

If you include the (SQL.90) or (SQL.100) version identifier in the URL you are taken to the specific version's copy of the topic. If you do not include the version identifier, you are taken to the copy of the topic in whatever is considered the most current shipping version of the product. Today that is SQL Server 2005. As part of the SQL Server 2008 RTM process (and I mean the real RTM, not the "launch"), MSDN/TechNet will flip a switch so the SQL 2008 BOL becomes the one you get with URLs that don't have a version identifier."

OLAP or Relational? Its all irrelevant

simonsabin — Fri, 04 May 2007 08:42:32 GMT

Chris Webb has mentioned an article about Ralph Kimball that discusses the what tool you should be using at the end of your BI process to present back to users. Should it be OLAP based or Relational based.

We had exactly the same discussion in the pub after the UG meeting last week. In my view its just data, OLAP is juts another big index and MDX a means by which you can represent a query of that index. Just like we have partitioning and windowing now in SQL 2005, why in the future can't we just have OLAP structures in the "database" in the future.

That way a user doesn't have to care what the structure is, they query using a tool and get the best response.

Ok so you still hav to design your OLAP structures, but its just another structure, we have to design tables, indexes, full text indexes, partitioning, XML indexes. Why not just have another structure?

PASS 2006 - Shared disk is not a scale out solution

simonsabin — Thu, 16 Nov 2006 10:31:31 GMT

At the end of the keynote today, Paul Flessner joined Steve Balmer for a short Q&A. One of the questions was around SQL Servers response to the Oracle RACK solution to provide a scale out solution.

Paul's response was that the Oracle solution is not a scale out solution because the disk is shared and unless you partition your application well you end up with contention over the global cache.

What Paul aluded to was that a scale out solution was originally planned for SQL Server 2005 but was canned, favouring the progamability (SQLCLR) of the server. However the work didn't stop. Paul mentioned that the preferred solution would build on a commodity server policy much like database mirroring. This is very much the architecture that is used for services such as Google and Terraserver, when many commodity servers are used to scale out and also provide redundancy.

So here's hoping that the solution, whatever it is, is delivered in Katmai.

CREATE or REPLACE feature request

simonsabin — Wed, 18 Oct 2006 23:59:54 GMT

I remember back when I moved from Oracle to 6.5 I found the lack of a create or replace really annoying. Whenever I create new sps I always curse at not being able to do "create or replace myproc..." having to do "if exists (select 1 information_schema...)" or "if object_id(...) is not null" or "if not exists(select 1 from sys.objects..)" is really annoying, especially as there are so many different options.

So I go to connect to put a suggestion and find one already there.

https://connect.microsoft.com/SQLServer/feedback/ViewFeedback.aspx?FeedbackID=127219

So if you want this implemented go and vote