Data Access Shakeout: What's A DB Developer To Do?

Microsoft's push, pointed out last week, is certainly not a revelation to those who follow it. But what should one who hasn't followed the machinations of this issue make of it? Or even more pointedly, what about someone who is moving to SQL Server and the .NET Framework?

Telerik CTO Stephen Forte recommends learning raw SQL, so if they use an object-relational modeling tool or either LINQ to SQL or the Entity Framework in the future, "they will know what is going on behind the scenes and use the raw SQL for the reporting solution as well as any complex queries and consider an ORM/LINQ/EF for the CRUD and simple stuff."

While Forte is concerned Microsoft's guidance on how to reconcile its various data access protocols won't be adequate for some time, he believes the shakeout will be organic. "Unfortunately the thing that makes Microsoft great and innovative, its sometimes disparate teams, leads to the confusion in the marketplace," Forte says.

In a blog posting of his own earlier this week , Forte pointed to a survey released by Data Direct Technologies last month that finds that 8.5 percent of.NET apps in production use LINQ to SQL as their primary data access method. "While this number is not huge, you can't ignore these developers voting with their feet by using LINQ to SQL in their applications," Forte says.

What's a LINQ to SQL developer to do? "Throw it all away and learn EF? Use nHibernate? No. The LINQ to SQL developer should continue to use LINQ to SQL for the time being. If the next version of the EF is compelling enough for a LINQ to SQL developer to move to EF, their investment in LINQ to SQL is transferrable to LINQ to Entities. If LINQ to SQL developers are to move in the future, Microsoft will have to provide a migration path, guidance and tools/wizards. (The EF team has started this process with some blog posts, but the effort has to be larger and more coordinated.)"

Microsoft will make sure LINQ to SQL continues to work in the .NET Framework 4.0 and will fix existing issues, wrote Damien Guard, a software development engineer in Microsoft's data programmability group (who works on both LINQ to SQL and the Entity Framework) in a blog posting in October during PDC.

"We will evolve LINQ to Entities to encompass the features and ease of use that people have come to expect from LINQ to SQL," Guard wrote. "In .NET 4.0 this already includes additional LINQ operators and better persistence-ignorance."

That's not to say new features won't shop up in LINQ-to-SQL, he added. "The communities around LINQ to SQL are a continuous source of ideas and we need to consider how they fit the minimalistic lightweight approach LINQ to SQL is already valued for."

Forte says LINQ to SQL developers will be ready to move to the Entity Framework when its feature set is a superset of the former and Microsoft offers migration wizards and tools for LINQ to SQL developers. "If Microsoft is serious about the Entity Framework being the preferred data access solution in .NET 4.0, [they will have to do a few things: "Make EF 2.0 rock solid. Duh. Explain to us why the EF is needed. What is the problem that the EF is solving? Why is EF a better solution to this problem? This is my big criticism of the EF team, the feedback I gave them at the EF Council meeting, is that they are under the assumption that 'build it they will come' and have not provided the compelling story as to why one should use EF. Make that case to us!"

Also, Forte is calling on Microsoft to engage with the LINQ to SQL, nHHibernate and stored procedures crowds.

Still there are many who are not happy with Microsoft's decision to give short shrift to LINQ to SQL, notably Oakleaf Systems' Roger Jennings, who last week said that LINQ to SQL will not go away simply because it is part of the current .NET Framework. Forte takes issue with that thinking.

"Just because something is in the framework is no guarantee that it will have a bright future," Forte said in his e-mail to me.

Jennings points to others who are weighing on this issue as well, such as Stu Smith, a developer at UK-based BinaryComponents Ltd.:

There's no one correct way to write an ORM. Different applications have different requirements. A general purpose ORM will never satisfy 100 percent of developers. Fine. I'm happy with that; there's a nice market for specialist providers.

What I'm not happy with is that while LINQ to SQL seemed to make 90 percent of developers happy, it's being replaced with LINQ to Entities that (judging by the feedback I've seen) makes far less developers happy.

I'm fine with the ADO.NET team writing a solution that fills that 10 percent gap or otherwise augments LINQ to SQL. I'm not happy with them replacing a 90 percent solution with a specialist 10 percent solution.

In the end, how this will all turn out remains to be seen, Forte points out. "We are still at the station buying tickets (to an unknown destination)."

What's your opinion? Drop me a line at jschwartz@1105media.com.

Posted by Jeffrey Schwartz on 12/10/2008