SQL Server 911 – vWorkshop now open for registration
Chris Shaw presents this in-depth (more than 4 hours!) virtual workshop about what you really need to know about SQL Server backup, restore, recovery and more. Learn what to do with specific, in-depth information about managing a failed restore, understanding all the options, what they mean for your installation and how to decide what you want to implement. Getting constant rave reviews, this vWorkshop also includes a certificate after you successfully complete the quiz, the slides, on-demand access, live Q&A with Chris during the vWorkshop and so much more.
[Register Here] (or even just check out the intro video and get the course outline)
Featured Article(s)
Accessing data distributed in SQL Server 2008 and EnterpriseDB
In this tutorial the idea of database linking native to MS Access will be used to show how you may run a query against data distributed on two servers, SQL Server 2008 and EnterpriseDB.
DBA School: Seats Are Already Filling Up
…and we’ve heard from a lot of you that are booking travel and such. If you’re coming to DBA School in September (3 days, Stephen Wynkoop and Chris Shaw teaching, in-person class), please be sure to register right away. We’re getting the message that there are a LOT of people that want to come. It’s pretty exciting the response we’ve already had. We’re hearing the price is unbelievable (accused of a typo in our pricing!), that the content is really strong and we just can’t wait to get started.
[ Register for DBA School here ]
I Need Your Help – and Feedback
(Survey closes Tomorrow)
I have a request. This week is survey week. I would absolutely LOVE to hear from you (yes, you) – it’s really important to get as many feedback elements as possible so we can see how we’re doing. I’ve split the survey into two parts – you can take the simple, 6 question intro survey and quit, or you can take both the intro and the more in-depth survey. I’d really love to hear what you think of features, what you use, what you don’t, where we should be going, and so much more. We’re giving away a SSWUG membership and a free pass to the fall Virtual Conference from a random drawing of people that fill out part 1. It’s only 6 questions. Please take a moment now and let us know how we’re doing…and, if you take part two, we’re adding more SSWUG memberships, more passes to the conference AND the conference passes will include the DVD set from the conference.
Optimizing your SQL Server Schemas
Derek writes in with thoughts: "I think this depends on a number of factors, including, of course, time and the knowledge of how your users will use the system. Often I have found that in a situation where system requirements are not completely know (is this ever the case?) then the easiest guideline is to model for the real world and worry about the reporting (and in some cases OLTP) aspects later. This may or may not require splitting OLTP and OLAP onto different databases. The advantage to "real-world" modeling is that it eases in understanding of the model and allows developers to ramp up quick as they can relate to the entities as they would deal with them (eg. Customers, Patients, Orders, Policies, Products, etc.) In some cases, reporting needs can be met on the OLTP db with proper indexes, materialized views or in some cases, using tools like QlikView which are completely geared towards NOT knowing up front what BI question you are trying to answer (I do not work for QlikTech ;)). I have often found that larger companies/systems that have reporting requirements are often comprised of various databases often on different technologies. In these case you need to have a point of consolidation regardless of how one application is designed – would this make the point moot?
In the end, IMO, the best design is one which makes the best use of resources and meets the needs of the users. If the requirements are known up front then a hybrid design is the way it go. When requirements are not know, save the time, design representing the real world and work out the details for OLAP (and perhaps OLTP) later."
There’s definitely something to be said for optimizing for the real-world. I also can’t tell you how many times "the real world" has not been known – no matter what we THOUGHT we knew – until after the system was really implemented. It’s a very key point. So many times we can solve a lot of performance issues by going back and reviewing an application to see how it’s *really* being used, vs. how we thought it might be used. Optimization can then begin…
Ralph’s thoughts on the subject were a bit more absolute: "Okay, let’s suppose that you use approach #1 (i.e. 2 schemas on one database instance or, even, within one database). So, now let’s also assume that you hit the month end reporting period and the month end billing period (both coming at the same time because, after all, you can’t schedule each month end at a different time). Now, what is the load factor for the database and what does this do for your OLTP processing when your OLAP loads up the server and the database?
Sorry, I have to go with Door Number 2 . . . separate the OLTP and the OLAP databases (and, therefore, the processing) entirely."
Agree? Disagree? Drop me a note – more tomorrow…
Featured White Paper(s)
SSIS Accelerator Series: Working with SSIS Expressions
Whether you’ve been involved with SQL Server Integration Services (SSIS) for years or are just learning SSIS, you’ve probably… (read more)