Instrumenting the Application

Posted by Kevin Merritt on March 31, 2008

Most of us, especially in the software industry, are data driven by nature. Given that propensity, I’m surprised by how few applications are purposely instrumented to provide real, hard data about how they are functioning.

This is one area where Microsoft does a very good job. Most Microsoft applications generate “Watson” data - information captured whenever an application has a problem. Subject to your personal privacy settings, that information is sent over the web to Microsoft so it can be analyzed, which hopefully results in the application being improved.

At blist, we’re doing similar things. We instrument the application to tell us things that we could potentially just ask you voice to voice. But we can produce better, more comprehensive and accurate answers simply by measuring it from within the application.

For example, we measure whenever anyone starts to import a CSV file but abandons the process before completion. We now know with certainty that we have more work to do in this area because we’re seeing that an unacceptable and unhealthy percentage of CSV imports are abandoned. Is the process too slow? Are we failing to provide enough feedback as the import progresses? Are we choking on big files? Are we having problems with the data in some kinds of files? These are the kinds of questions that can best be answered by instrumenting the application to tell us with certainty. And that’s exactly what we are doing. Look for the CSV import process to improve over the upcoming weeks as we analyze the operational data our application generates.

Share and Enjoy: These icons link to social bookmarking sites where readers can share and discover new web pages.
  • Digg
  • Sphinn
  • del.icio.us
  • Facebook
  • Mixx
  • Google
/

3 Responses to “ Instrumenting the Application ”

[...] 31, 2008 Software Tags: Instrumentation blist, Kevin Merritt Kevin Merritt over at blist has a post up on software instrumentation — aka “developer crack” because most developers LOVE to write these types of [...]

Left by Instrumentation is Developer Crack « The Pursuit of a Life on March 31st, 2008


Don’t know exactly where to start, but your first comment about Microsoft is the main reason I abhor anything they do anymore. Having used and programmed in the Windows environment for the better part of 25 years, I can’t begin to tell you how many hours of wasted time the OS has caused in data corruption, non-date file corruption, hard disk crashes, malware and virus attacks, and, well, you know all the other pitfalls inherent in their software. Then there’s all the time and dollars it’s cost me in down time for work in process for my own clientele.

And all this because Microsoft does a “good job” at collecting Watson-data? Uh, what are you smoking? If they were so good at it, why then are most users if given the alternative reverting back to Windows XP SP-2 and the upcoming SP-3 rather than Microsoft’s Vista debacle?

But enough ranting on this first issue.

I think Blist has some interesting capabilities and I also think the future looks pretty bright for the product. However, I think you’re going to need to somehow make it relational in the sense that you can not only design around it, but also import into it from such databases as MySQL, Filemaker, etc., in such a way that primary keys can be set up and sustained in this data merge. Once into Blist, related blists are handled as they are now, but with much greater latitude.

For example, under your current design limitations it would seem quite difficult to set up an inventory control system or invoicing system where so many parent and child data fields would reside on one screen and where updating/creating/cascade deleting the records could take place in a transaction-type environment.

I do realize that right now this isn’t the intent of the online/browser based app, but it is something that more and more people are going to ask for shortly as they try to push the envelope with what you currently offer in terms of feature set.

Still and all, it’s a very nice interface thus far, has a some ways to go, and I’ll be interested to see and use it as it evolves.

Left by Bart Bartholomay on April 1st, 2008


I’ve tried to upload a .csv a number of times. But never had success. It just doesn’t do anything! Maybe it doesn’t like macs, or something. Maybe Safari? Firefox? Anyway. That’s my reason.
Once I can upload csvs I’ll be blisting to my hearts content.

Left by Peter on April 6th, 2008


Something to say?