Microsoft took many by surprise when it announced in a blog post on June 23 that it is not pursuing the previously planned separate delivery of WinFS or the second beta for the product.
Corey Thomas, the group product manager for SQL Server, explained to eWEEK Senior Editor Peter Galli the rationale behind this decision and why Microsoft believes it is good for most of its customers, partners and developers.
Can you explain the enormous discrepancy between all the information about WinFS that was shared just two weeks ago at Microsofts TechEd conference, including talk of a second beta for the product, and this latest announcement that you are not pursuing a separate delivery of WinFS, including the previously planned Beta 2?
The presenters at TechEd were operating in good faith. The decision to do this was actually only made late last week. Even though we had major discussions at TechEd, we wanted to get this latest information out to our customers and partners as soon as possible, even though we knew we would take a bit of a knock for having one conversation at TechEd and then announcing the changes in the ship vehicle a couple of weeks later.
So, what you are saying is that the presenters at TechEd knew these changes were under consideration but that no final decision had been made?
The decision was only made after TechEd. But everyone has always known that SQL Server was part of the data platform strategy and they were working closely with the SQL Server team and that has been an ongoing discussion.
to read more about what Microsoft had to about the future of WinFS at TechEd 2006.
What they didnt know was that a couple of weeks later we would make a decision to change the ship vehicle strategy. That decision was only made late last week.
Tell me why that decision was made and what the basis in thinking is for it.
We have talked a lot recently about our data platform vision strategy and one of the big things that Paul Flessner [Microsofts senior vice president for server applications] had started talking about a couple of months ago was our data platform vision and that had a couple of pillars behind it. One of those was the idea of all data and going beyond just relational data and helping customers get more value from their structured and unstructured information assets. We also looked at other things around cost, complexity, TCO and business intelligence.
The thing that shocked us a little bit was the overwhelming positive response we got from customers to the idea of how best to manage this growing volume of structured and unstructured data. As we took a harder look at the feedback and how best to bring this to market and execute on it, it became increasingly clear that leveraging a lot of the more mature incubation technologies inside of WinFS was going to help us deliver on that promise and goal.
Directions and Roadmaps
So, what you are saying is that customers told you they would rather see you productizing the mature aspects of the WinFS project into SQL and ADO.NET at the expense of Windows and Office, right?
We havent commented on the Windows direction and roadmap post Vista. Right now we are not talking about post-Vista plans for Windows and we are only saying that these technologies will be included in SQL Server and that we will not have the monolithic software component of WinFS in beta 2 form. That does not preclude us from using these technologies in Windows going forward.
But one of the things the Windows team has said is that WinFS would be made available as an add-on subsequent to the release of Windows Vista. So how will these technologies now find their way into Windows going forward?
Well, to be clear, there will now not be a WinFS add-on subsequent to the release of Vista. But we are still focused on the integrated storage vision for SQL Server, Windows and everyone else. That is what we are trying to accomplish; the end goal. But what we have announced is how this will happen with SQL Server, but we are not talking about how that will happen for Windows post-Vista, although I will say that we delivered a good amount of the original vision inside Vista, but we are not talking about what the next version of Windows will look like.
So are you saying that the goal of a single, integrated store for Windows, Office and SQL Server data is still alive and well?
The integrated storage goals are still there, alive and well. We have decided not to have a stand-alone software component that shipped post-Vista, but those goals are still there. But we are still real early in the planning cycle and are not talking about that.
So, to be clear, there is still the possibility of a single, integrated store for Windows and Office and SQL down the line?
Yes, Windows could leverage the technologies that are in WinFS in future versions.
There is a lot of unhappiness about this decision as many people wanted to see Microsoft integrate the relational file system with the Win32 API. They are now concerned that this is not going to happen at all. Is that the case?
That is an incorrect assumption as we have only talked about one aspect of our vision around the data platform and we have received positive feedback around that. A lot of people are asking about the roadmap going forward, and we are just not discussing that at this point in time. We are really focused on making sure we deliver Vista, which we think is a great release, and which delivers a lot of the WinFS experience we first talked about with regard to search and organization. There is some speculation as people dont know and we are not in a position to disclose more as we are focused on Vista right now.
What components of WinFS are in Windows Vista?
In terms of the goals of integrated storage, we delivered on a good number of those in Vista, especially for the end users. I think the search experience in Vista and the ability it brings to better manage information were core to what we talked about and are in Vista.
If you separate out the different constituencies, end users will have an amazing experience with Vista around search and the organization of information. And, for the programmers, we actually have some pretty robust programming stuff with the WinFX and .Net stuff that will be in the next version of .Net. But there is some angst about when the pure relational file system will show up and what is going to happen there.
There was a lot of talk at TechEd about how all the data generated by Windows Live services was complicating the idea of WinFS as a single relational store. Was that a contributing factor to move in the current direction?
The primary contributor to our decision was how best to enable the data platform stuff as embodied in SQL Server, so that was a primary driver and designed to meet customer needs. We get so much feedback about the struggle between structured and unstructured data and how they can be tied together as well. Customers have all this data in their database and all this information in files, all of which is actionable and critical to their systems and they want to know how they can bridge those together.
How will Windows Live services now be able to take advantage of the WinFS technologies?
This is something we are still working on. We think there are good opportunities in the information space, but this is something that is in process and that we are still looking at.
There is talk that this decision was in part based on the fact that the WinFS code was simply not up to scratch and the team needed more time to work on it rather than admitting beta two would take longer to be released than planned. Is there any validity to this?
Not really. You have to step through the decision-making process: Once we got the feedback and saw that there is a real customer pain-point in managing the data that is out there, then we have to look at the most impactful, large-scale way we can address that, and then you have to prioritize to deliver on that.
Given the huge role SharePoint is playing in Microsofts future, some say we will see many of the features that would have been found in the stand-alone WinFS show up there. Is that a fair assessment?
Well, SharePoint runs on top of SQL Server and there is lots of synergy between those two teams.
But we look at SharePoint and SQL as a package of things together, and it is premature to talk about specific features being found in one rather than the other.