Do you plan on upgrading to Microsoft SQL Server 2008? Well, this has certainly been a hot topic for our readers. We’ve had an overwhelming response to an editorial and commentary on this subject. In fact, we’ve devoted an entire page in the October issue to give you an idea of where our readers stand. But, like every opinion, there is always another opinion that has merit and is worth considering. In response to “Too Soon for SQL Server 2008”, one of our regular forum posters and SQL Magazine reader, Kalman Toth, sent us his opinion on why he’s taking a serious look at SQL Server 2008. Kalman has been a regular contributor on our forums since 2005. His article follows:
Is SQL Server 2008 Necessary? — Kalman Toth
When I was at the SQL Server 2005 introductory conference in Las Vegas, Microsoft executives promised faster version release as demanded by “us”, the user community, no more waiting for 5 years (since SQL Server 2000). When I was at the Microsoft Business Intelligence conference in Seattle this May, Katmai was announced. A month later it was gone. It became SQL Server 2008. That is fast! Informally I asked several attendees if they converted to SQL Server 2005. Most of them were in the middle of upgrade.
Personally I believe that the 5 year software release cycle is an excellent one from the business viewpoint of database software customers. Software vendors have to prepare a new release of their software to take advantage of the new features of the latest SQL Server. That can take a year or two easy. IT staff has to be trained and become proficient using the new features. Careful migration plans prepared and carried out. Last but not least, companies want to get high ROI on SQL Server. A CTO does not want to disturb a well functioning system with an upgrade for no good reason.
So who are the people who are asking Microsoft for new features and say that 5 year is too long to wait? Among them I. No-blocking database re-indexing is very important in 24/7 operation. SQL Server 2000 did not have it. SQL Server 2005 has it. 5 years really appear to be an eternity if you are a DBA for a high-volume website with a SQL Server backend. Table partitioning probably was on the wish list of most SQL Server 2000 installations. SQL Server 2005 has table partitioning after 5 years of long wait. Unfortunately it is not exactly automatic; it requires elaborate setup and maintenance. For example, it does not support “rolling” partition. You cannot partition the last 90 days of data from data older than 90 days, a very natural requirement when partitioning data based on acquisition date. We can go through all the great new features of SQL Server 2005 and surely we will find a segment of the user community who wanted it and feel that 5 years was too long of a wait. Features like: XML processing, tree processing, try/catch error trapping, snapshot isolation, MARS, CTE, deadlock visualization, ROW_NUMBER function, non-key columns in indexes, column-level encryption, database mirroring and so on.
But then again there are features the user community is waiting for and still not in the 5 year cycle release of SQL Server. I am waiting over 15 years for example a DATE (only) datatype. In most business applications, the time part of datetime datatype is not required. Birth date, employment date, publication date and deposit date are all examples for it. I am also waiting for a long time for a database backup file compression and encryption utility. The fact is that many of the missed features are already on the plate of Microsoft development managers responsible for SQL Server. Yet they have to make decision what to include what to postpone in the next release of the software.
There are of course the surprises you wish did not come with the new version. I liked Enterprise Manager of SQL Server 2000 very much. I did not have any trouble with it. I also liked DTS. I liked the stored procedure debugger in Query Analyzer. Well, they are all gone, replaced with something better or worse, but certainly not equivalent. For most DBA-s, it is not an easy transition from the practical DTS to the sophisticated ETL tool SSIS. For most SQL Server shops, SSIS may be overkill. Even more astounding for the average DBA, that Microsoft does not even call SSIS a data transfer utility, rather a Business Intelligence tool for data warehouse population which is removed from the DBA’s command control tool: Management Studio. Life is of course full with compromises. That is the price of progress. All the great features of SQL Server 2005 weigh more on the scale than the negative surprises.
So why to have SQL Server 2008 and not SQL Server 2010? What are the new features which we are waiting so desperately? New date/time types for one thing. Nobody can argue with these especially if they include the DATE (only) datatype. Date is the bread and butter of most database application. SQL Server 2008 has a new geometric (spatial) datatype which can be used for satellite tracking and map applications for example. I am not convinced about this one. Companies who need this feature have already acquired the appropriate application or Business Intelligence software. Some of the new features are just bettering the features of SQL Server 2005. As such they could be released with Service Packs. Here is one new feature I really get excited about and vote not to wait till 2010: Performance Studio for troubleshooting, tuning and monitoring servers across the enterprise. PS is every DBA’s dream who ever struggled with optimizing slow-running queries, finding the right indexes and eliminating performance bottlenecks. There is also the Pervasive Insight feature which will not excite most DBAs and developers due to lack of understanding. Microsoft’s definition: “Drive business intelligence throughout the organization with reports and analysis of any size or complexity”.
Microsoft’s overwhelming corporate is objective is to get a piece of the fast growing Business Intelligence market. That was stated clearly and unequivocally at the May BI conference by Steve Ballmer, CEO. Microsoft sees SQL Server not as a database product anymore, rather just a backend tool in its Business Intelligence offering. To achieve its long-term goal of gaining market share from Oracle, IBM and SAP in BI, it does make sense to have a 2-3 year release cycle for SQL Server and associated BI software like Analysis Services, Integration Services and Reporting Services. From the standpoint of the average database software customer, 5 years cycle with improvements in Service Packs is not a bad one. From the standpoint of DBA, developer, BI developer and database architect, faster release cycle is the better. After all, IT knowledge workers welcome new knowledge and higher software development productivity.