I have been meaning to share the results of this survey for quite a while.  It has been been one of those things that kept dropping off the pile of todo items until now.  I am going to show the summaries and then give some of my observations about the open ended questions.

What survey?

This survey was taken by over 200 VistaDB users during October / November 2007 (wow, has it been that long?).  Most of the questions were multiple choice, but some had fill in the blank type answers because I wanted honest user feedback.  None of the users gave their personal information other than when answering if they were a current customer or trial user.

Overall, how would you rate VistaDB?

• Good or Excellent  78%
• Average 9%

What feature(s) attracted you to VistaDB?

SQL Server syntax / SQL Server datatypes	12%
CLR Procs / CLR Triggers – Advanced database features	4%
ADO.NET 2 Provider	13%
Direct Data Access (DDA) – not having to use SQL	7%
Deployment options (LAN, Local file, Isolated Storage)	11%
In process database for your application	11%
XCopy deployment ability – no complex installation	13%
Shared hosting capability for ASP.NET websites	4%
Fully managed database engine	15%
Source code availability	3%
Mobile Application Support (PocketPC / Windows Mobile)	4%

The only thing I will comment about here is that I find it interesting how scattered the results were.  Obviously some of the more whiz bang features were ho hum.

Did you recently download the VistaDB Trial?

• Yes 41%
• No 59%

During your recent trial of VistaDB did you:

• Installed, but never tested  8%
• Installed and tested  82%

If you did not purchase VistaDB, what were your reasons for deciding not to buy:

• Performance 25%

The rest of the answers were scattered all over and were insignificant in percentages.  I am not sure if performance is the real issue, or the expectation of performance.  It appears that some users expect us to run faster than SQL Server, or think SQL Server is slow(!).  C# code in a managed environment versus C++ code with 10+ years of optimization is not a comparison I want to make.

Did you read the initial readme after installation?

• Yes 66%
• No 33%

Did you follow the Getting Started Tutorial link in the readme and watch the online presentation?

• Yes 55%
• No 31%
• Didn't know about it 14%

Did you try the example solutions loaded with your trial?

• Yes 51%
• No 37%
• Had problems 9%

How could the samples be improved?

• Add samples projects that demonstrate more complex topics  30%

The other answers to this were scattered.  Basically we got it that people want more complicated examples showing advanced features, and best practices.  It is incredibly difficult to construct meaningful examples like this though without losing the intent of the example into code practices / tier design / methodology wars.

Did you run the NUnit examples included with VistaDB?

• Yes 10%
• No 76%
• What is NUnit? 14%

Did you examine the NUnit source code for code snippets or ideas for how to use features?

• Yes 17%
• No 83%

Obviously NUnit is much more about regression testing for us than it was for this base of users.  I was a little surprised considering how many complaints I had when I took over about quality control and regression.  But I guess users assume we are doing our job now and taking our word for the NUnit tests.

What other features do you need VistaDB to add?

This was a fill-in answer, and wow did I get an earful on this topic.  Everything from implement the impossible (VDB3 in VB6, add ISAM indexes, rewrite in C++, etc), and all sorts of statements for us to add things we already support (Transactions, Views, FTS, etc).  We need to do a better job of documenting and explaining what we already provide to users.

Do you visit the online forums?

• Yes  85%
• No 15%

Do you participate in the forums?

• Yes 21%
• No 79%

Have you filed a report through FogBugz?

• Yes 19%
• No 81%

How would you rate the current help system?

• Poor 18%
• Ok 55%
• Good 26%

What topics would you like to see in the help files?

• More code / low level examples  35%

The other choices were mixed about even.  Help files tend to be a lot more useful when they contain clear, concise example code of the feature.

Should the VistaDB help include a complete SQL Syntax guide with examples of how to call the supported SQL Functions?

• Yes  77%
• No 23%

What other database engines do you currently use in your products?

• SQL Server 2005 37%
• Access MDB 20%

There were other options, but these two were by far the leaders.  It is to be expected considering our product feature set and API that most of our users would come from a Microsoft centric product. 

What SQL Server features are we missing for your product today?

This was another fill in the blank that resulted in huge variations.  By far the largest request was TSQL Procs in some form (which we now have in 3.3).  The other valuable feedback was the number of users who said they prefer to avoid using SQL Server specific features in favor of more standard SQL to avoid vendor lockin or version problems in the future.

The other questions were about advertising and where users read dot net information.

What did we learn from the survey?

Features are good

The most important lesson was that the feature set is pretty much right on the mark.  Users like what they have, but don't always understand it due to a lack of documentation and samples.  That is probably a pretty fair statement as well.  We do not have any full-time documentation people, or technical support people writing samples.  It is an area that we will continue to improve in the future.

Managed Code

Our user base seems split between those who really care and are passionate about managed code, and those who could care less and just want a cheaper SQL Server.  That was a little surprising.  The knowledge of how managed code works and why it is different was not very good, but the SQL database knowledge was quite high.  It is interesting.  I can remember when a certain object database vendor first released their object database and a lot of people rushed to compare bulk insert operations per second and slammed the database as too slow and not up to the standards of other SQL databases.  At the time I didn't grasp that they are truly different (written in Java I think).  Since that time they have taken great pans to avoid the SQL comparisons and now I understand why, they really are not valid or important. 

Tools make the difference

In one of the fill in sections about half of the users took the time to comment on the Data Builder and what a great tool it is.  Many cited it as the reason why they chose VistaDB over SQL CE ("The XCopy deployable Data Builder for working on site with a client database is priceless. SQL CE requires I copy the database to my laptop or have much more complicated tools installed on the client machine.").  I don't think that many users understood that we include the source to the Data Builder will all registered licenses.  It is probably the ultimate code sample, entirely written in C# and customers have the code to play around.