In-Depth
SQL Server 2008: A Case of Too Much Too Soon?
Are customers -- many of whom are still satisfied with SQL Server 2000 -- interested in an upgrade to Microsoft's SQL Server 2008?
Last week, Microsoft Corp. announced its latest incarnation of Windows: Windows Server 2008. Included in the launch, the company touted its upcoming SQL Server 2008 release, slated to ship sometime in Q3 of this year.
If it seems like only yesterday that Microsoft was trumpeting a new SQL Server revision, that's because, comparatively speaking, it was only yesterday: Microsoft officially unveiled SQL Server 2005 two years ago last November -- more than five years after it shipped its venerable SQL Server 2000 database.
If Redmond hews to its product schedule -- a big "if" -- it'll deliver SQL Server 2008 just three years after its last major SQL Server overhaul. Which begs a big question: will customers -- many of whom are still satisfied with SQL Server 2000 -- be ready for an upgrade?
In contrast to SQL Server 2005, next-gen SQL Server doesn't pack much of a business intelligence (BI) wallop.
SQL Server 2005, you'll recall, introduced a revamped data integration facility -- dubbed SQL Server Integration Services (SSIS) -- which supplanted Microsoft's existing Data Transformation Services (DTS) technology. SQL Server 2005 also introduced improved Analysis Services (AS) and Reporting Services (RS) components, as well as a new developer-friendly -- and, according to skeptics, data management-threatening -- programming facility (see http://www.adtmag.com/article.aspx?id=11148&page).
SQL Server 2008, by contrast, does ship with improved unstructured (and semi-structured) data=handling capabilities, as well as significant support for geospatial data types. Both enhancements should be of interest to BI pros -- but neither is as significant as SQL Server 2005's SSIS refitting.
There are a couple of ways to look at the issue, say SQL Server pros familiar with Microsoft's next-generation database. For customers who haven't yet made the leap to SQL Server 2005 -- and there are a lot of them (including a number of customers who are still running on SQL Server 7.0) -- next-gen SQL Server might provide an attractive upgrade path.
"I have clients on both [SQL Server 2000 and SQL Server 2005] -- probably 50/50 currently. The main reasons for not upgrading are application compatibility and hardware replacement cycles," observes SQL Server MVP Kevin Boles, a principal with Indicium Resources Inc. "I have no clients currently pursuing immediate upgrade to SQL 2008, [but] I do have several [who] will benefit greatly from features in the new version, and I will be encouraging them to evaluate it sooner rather than later."
SQL Server MVP Andrew Kelly, for his part, says most of his customers -- who tend to have very large SQL Server implementations -- are already running on SQL Server 2005, which has clear advantages over its predecessor. For these customers, Kelly argues, SQL Server 2008 amounts to a logical upgrade path.
"I do feel a lot of small- to mid-sized companies will hold off and go straight to 2008 if they haven't upgraded already. If they didn't absolutely need any of the features of 2005, they can most likely afford to wait until 2008 and have more time to prepare for the upgrade," he says.
"Many of my clients are high-end and as such most have already switched to 2005 by now. I have only dealt with four clients in the past two years that were either in the process of upgrading or were still on [SQL Server] 2000. All others have already made the move."
For large shops, Kelly continues, SQL Server 2008 should be hard to beat. "For high-end or larger systems, there are a lot of advantages of using 2005 over 2000. SQL Server 2008 will add to the features for the enterprise, and I feel they will be adopted by many of [my clients] as well once it is released. I have already worked with one very large customer who will be running a pre-release version of 2008 in production very soon. From what I have seen of the product so far, I have been impressed with the stability of the platform at this stage in the game."
Other SQL Server pros aren't quite so sure. Even shops that are heavily invested in Microsoft's BI stack -- and that could benefit from SQL Server 2008's improved support for unstructured and semi-structured data -- are hesitant about immediately taking the plunge.
"It is likely we will switch, although I will have to make a business case for it to happen [so] shortly after release," says SQL Server DBA Gregory Beamer, who asked that his company not be named. Beamer's employer is up and running on SQL Server 2005, which addresses most of its needs.
"Thus far, there are not a huge number of compelling features that we need," Beamer continues, citing SQL Server 2008 amenities such as transparent data encryption and improved BLOB support as worthwhile improvements. "I will be assessing SQL 2008 more thoroughly as we get closer to release to determine whether it is worth switching to this year. I imagine the answer to be no, however."
Beamer's experience differs from that of another SQL Server DBA, James Underwood, who likewise requested that his company not be named.
In Underwood's case, his employer is still running on SQL Server 2000. Outside of a single SQL Server 2005 deployment -- largely to take advantage of that product's improved SQL Server Reporting Services facility (as an alternative to Crystal Reports) -- his organization is getting by with its SQL 2000 status quo.
"We have not upgraded to 2005 for budgetary and time -- [e.g.,] upgrade testing, manpower -- constraints. We do have one small … server set up for SQL Server [2005] Reporting Services," he indicates. "I think there is a chance we will go directly from 2000 to 2008 in 2009 or 2010. If we don't stay on 2000, then it will make the most sense to skip 2005 altogether, since upgrading two versions [i.e., skipping an interim release] is usually not much more work than upgrading one version."
While SQL Server pros seem satisfied with that platform, they do admit to a few concerns about Microsoft's development roadmap. Instead of fast-tracking its SQL Server 2008 development efforts, SQL pros say, Redmond might want to focus on improving its existing SQL Server 2005 product.
"I am mostly happy with Microsoft's SQL Server road map and definitely happy with the way they are growing the market," stresses SQL Server MVP Boles. "One glaring deficiency is the lack of commitment to producing a Service Pack 3 for SQL 2005. Many users out there agree with this based on various forum postings and blogs. Judging from the quality of the 2005 SP2 release, they need to pick up the testing and quality a notch as well. Actually it seems that quality has been not quite up top par going all the way back to SQL 2000 SP3."
Analysis Services (AS) authority Chris Webb, a principal with UK-based Crossjoin Consulting, likewise flags the absence of a SQL Server 2005 SP3 release.
"A lot of people have complained about this, especially because SP2 RTM contained so many bugs and although these bugs have been mostly fixed in subsequent cumulative update releases, most companies don't install anything other than service packs," he points out. "I've heard, though, that Microsoft might be about to make a U-turn on this decision since it provoked such an outcry."
SQL DBA Beamer says he's quite happy with Microsoft's SQL Server development effort. It's the rest of the Microsoft juggernaut that frustrates him.
"I am not overall thrilled with Microsoft at this juncture. Currently, the speed of technology advancement is far outpacing the tool support. What good is a new technology if you must lock all of your 'rock star' developers into that technology to utilize it?" he says.
"This is not aimed at SQL Server, in particular, as the SQL Team seems to wait a bit over some other teams, but it does apply strongly to Silverlight, the Entity Framework, many of the WCF, WF, and WPF advancements, [and] ASP.NET Extensions," Beamer continues.
"On SQL Server, in particular, the product has gotten better every release since 6.5 [circa 1996]. SQL [Server] 2000 was a monumental release, and SQL 2005 builds on the foundation. From what little playing I have done with SQL 2008 -- during the Katmai early beta phase -- it is a great improvement, although most of the improvements are not in the arena of 'new features,' per se."
Other users, such as SQL DBA Underwood, wish that Microsoft would sharpen its data management focus -- that is, concentrate on improving SQL Server-the-relational-database, instead of building in add-ons, extra bells and whistles, and developer- or manageability-friendly amenities.
"In many cases I think they are building too much into their products: Office, SQL Server, Windows, all of them," he says. "There are core things that each product is good at and they tend to become over-engineered, bloated, and unwieldy when they try to put everything into one package. I think the focus should be on performance tuning, ANSI SQL extensions, backups, and recovery, the things a RDB[MS] are supposed to do."
In the final analysis, says Crossjoin Consulting's Web, SQL Server 2008 is an evolutionary -- rather than revolutionary -- release. For this reason, he suggests, few users -- outside of those stymied by long-standing SQL Server 2005 bugs -- are considering moving to it immediately, and this is as it should be.
"Almost no one is considering migrating to [SQL Server] 2008 at the moment; the only exceptions are those who have come across bugs that will only be fixed in the 2008 release or who have severe performance problems and expect to benefit from the improvements in the AS calculation engine that have been made," he comments.
"These companies are in a very, very small minority though. Apart from the performance improvements, all of the other new features in [Analysis Services] 2008 are just 'nice-to-haves' which make life easier but don't really allow you to do anything that you couldn't do before," Webb continues. "This is in part intentional, because the changes between [Analysis Services in SQL Server 2000] and [Analysis Services in SQL Server 2005] were so great that the dev team felt that more upheaval would be a bad thing."