Heres the joke in Oracle Land: Whats the proper way to manage SQL Server performance disruptions? Answer: Turn it off.
While Microsofts DBMS has done some slick work recently on TPC benchmarks, its still the butt of jokes like that to many Oracle pros. And you have to admit, it has deserved snickers on the basis of reliability and security. But is that attitude still warranted, or, given recent SQL Server gains, are we talking about a perception problem?
Perhaps the latter. With Tuesdays release of Winter Corp.s surveys of the largest and most heavily used databases, Microsoft for the first time ever can boast that SQL Server on Windows made the Top 10 list for biggest OLTP databases. Verizon Communications weighed in with a 5.3-terabyte SQL Server transaction-processing database, reaching sixth place in size for all environments, the top transaction-processing database on the Windows platform.
While scalability has certainly been a concern for SQL Server, as Winter President Richard Winter noted to eWEEK.com, Verizons showing in the Top 10 demonstrates that progress has been made.
Thats important. As anybody will tell you who works with todays ballooning databases, there are three things that matter: performance, performance and performance. "Its pretty straightforward, because with large databases, data sets and volumes are becoming massive," said Ian Abramson, who is CTO at Red Sky Data Inc., in Toronto, as well as co-author of several Oracle books, the current director of education programming for the International Oracle User Group, and an expert on Oracle data warehouses and applications. "[Those large databases] need to analyze a lot of data over a long period of time," he said.
Telecoms need to analyze potentially millions of telephone-call records per day. Its crucial that, as databases analyze such extremely large data sets, they grant flexibility in looking at the atomic level of data—i.e., before aggregation occurs. Its also crucial that data can be analyzed on the fly, with real-time analysis, to determine trending. Whats paramount is that all this gets done fast, fast, fast.
Scalability is key to get through those millions of records, and Microsoft has made progress there. Still more progress is on tap in the upcoming SQL Server upgrade, code-named Yukon. Reliability should also see a boost with new continuous-availability features that are geared to minimize scheduled downtime, according to Stan Sorensen, director of SQL Server product management. A new online index-rebuild feature is designed to reduce the need for DBAs to take the database offline as well.
But regardless of such current or upcoming gains, Microsoft still has to prove to large enterprises that the company "gets" large enterprises. While reliability gains are reducing the need to shut SQL Server servers down and then power them back up when things arent working right, the perception remains that reliability just isnt there yet. As pointed out by Charlie Garry, senior program director at Meta Group, security is the other monster under the bed, and Microsoft hasnt slain it yet. "The security issues with the viruses and attacks and so forth are obviously significant," Garry said. "The ones who say they wouldnt touch SQL Server, thats what they point to as an issue."
Regardless of security, reliability and performance issues, however, SQL Server is sneaking into organizations—and thats perhaps the most serious problem of all, according to Garry. "Lets face it, the majority of people running SQL Server-based applications are not doing so in an organized, professional manner," he said. "Theyre running it on servers under somebodys desk. Ive been in shops where Ive seen SQL Server database servers [with] Coke cans on top of the boxes."
While Microsoft still has work to do with reliability and security issues, much can be solved were organizations to start to take SQL Server more seriously. That means treating Windows platforms and SQL Server software as important components of the data center, bringing them out from under the desks and into the data center, and subjecting them to the same processes and procedures that are in place for Oracle and other important environments.
That means paying attention to change management, backup and recovery, high availability, the whole nine yards. With that kind of respectful treatment, SQL Server will deliver better reliability, and, finally, once the rings from the Coke cans are wiped clean, it might have a chance to grow up and play with the big boys.
Discuss This in the eWEEK Forum
eWEEK.com Database Center Editor Lisa Vaas has been writing about enterprise applications since 1997.