Bill Graziano, VP of marketing for PASS, was first on the
stage at todays keynote. He said that he would provide the shortest keynote
address of the week and I think he met that promise. Bill recognized the
current board members, and introduced the new board members. He then invited
Wayne Snyder on stage to present an award.
Wayne said PASS wanted to recognize one of those long-term
volunteers, Kevin Kline. Kevin has supported PASS in many aspects for the last
ten years. He is the current past president of PASS. Kevins award
presentation got a standing ovation.
Bill Graziano finished his address by announcing the
location and date for next years PASS summit. It is once again coming back to
Seattle. It will be held from November 8th through the 11th.
It is expected that next year Microsoft will announce a new product release for
SQL Server. John than introduced the next presenter Patrick Ortiz from DELL.
Patrick is the Practice Manager for SQL Server and Business
Intelligence (BI). Patricks talk was on Configuration Management and how it
helps with Disaster Recovery and Consolidation. Bill stressed how important it
is to really understand your SQL Server environment, so that you can leverage
this information to make sure you develop the appropriate strategy for your
Disaster Recovery and Consolidation projects.
Patrick first explored Configuration Management. The key
components of Configuration Management are to identify the assets you have,
like instances, databases, servers, etc. For each asset, you need to identify
how critical the asset is to your business. Some assets are critical, like
your production servers, where as a test server might not be as critical.
Patrick then explained some of the benefits of doing
Configuration Management assessments of your SQL Server assets. By doing
Configuration Management, you provide a clear picture of the SQL server
environment you have. Understanding your environment allows you to define
priorities based on criticality. These priorities will help drive your
Disaster Recovery and Service consolidation efforts.
Patrick finished off his presentation by talking about the
key steps to performing a Disaster Recovery and Consolidation effort. He said
each Disaster Recovery or Consolidation effort needs to take on three different
steps: Planning, Implementation and Maintenance. Your Configuration Management
efforts provide you the key information that you can use in the planning steps
to determine the architecture needed to support your target Disaster Recovery
or Consolidation environment. Once you have architected the environment you
then need to implement the solutions. Finally, after your Disaster Recovery,
or consolidated environment is up and running you need to maintain it and
develop a performance baseline. By running these performance baselines over time,
you can determine the hot spots that require attention.
Last up was a presentation from Dr. David DeWitt. David
talked about technology trends from 1 to 1000 MIPS. David runs the Microsoft
Jim Gray Systems lab in Madison Wisconsin. He has been working on release 1
and 2 of the SQL Server Parallel Data Warehouse. His presentation for todays
event was to look at the history of databases over the last 30 years, and how
technology changes have improved and hindered what we do today.
Dr. David DeWitt giving his 1 to 1000 MIP presentation
David discussed how over the past 30 years hardware has come
a long way. Below are some statistics he had. These statistics compare how
much faster or bigger things are today verses how they were in 1980:
CPUs are now 2,000 times faster
CPU cache is now 1,000 times faster
Memory is now 1,000 times bigger per CPU
Disk drives are now 10,000 times bigger
David says the speed of the CPU is way out of whack with the
disk capacity of todays disk drives. This lopsided speed verses capacity
makes it very hard to optimize performance. Solid state and phase change
memory will really change our data storage life in the future.
Today we store data as a row store. Meaning all columns of
a records are stored together in a row. But when querying data we typically
are searching and returning data at the column level. We need to start
thinking about how we can read as little data as possible to answer our
business query. Data stored in the row store doesnt fit into the model of
minimizing the data read when we retrieve data based on columns. Database
architecture of the future needs to consider column stores. A column store is where
row data is stripped into each individual column and the column data is stored
together in a tightly compressed format. David said that database management
systems of the future will most likely store data in a row store, or column
store, or both. This will facilitate fast updates for row stores and fast data
retrieval for column stores.
David ended his talk by saying he couldnt exactly state the
direction Microsoft is planning on heading. He did say if you followed his
talk on column stores and row stores, and the hybrid of both types you might be
able to figure the direction that Microsoft might be heading. Nevertheless,
he was able to give one firm direction that Microsoft is heading. He stated
that Microsoft will be shipping VertiPaq with SQL Server 10.5. VertiPag will
be an in-memory column store.
This article ends my series of the daily updates around the
PASS Summit 2009 keynote addresses. I hope that this series provided help on
Microsofts directions and announcements for those of you who were not able to
attend PASS. My take away from all this is I should probably download the
November CTP of SQL Server 2008 R2 and PowerPivot, then play with this new
functionality. Doing this will allow me to start thinking about how this new
version of SQL Server and power user tool can be leveraged to shape my
organization's future Database and Business Intelligence direction.
See All Articles by Columnist Gregory A. Larsen