This is all speculative at this stage but assuming I opted for a web application rather than a traditional desktop approach, does anyone have any thoughts on data management?
Copying the training.mdb and the (new) log files each time isn't very efficient but it would add a degree of security to the user - assuming that they were recoverable. You could download to any PC if you were away from home or re-instate your data without needing a dedicated backup strategy. You could improve performance and reduce transmission costs by sending deltas.
From my perspective, the downsides cover a couple of areas. There's data segregation - ok, this isn't SSN's or credit card info but there's some danger of data leakage and building robust web applications isn't as easy as all that (my day job is IT security focused in the banking industry, so I've seen some absolute horrors over the years!). There's the effort of a direction change.
Finally, there's cost - storage costs, there's no two ways around it. Services like trainingpeaks are operated on a commercial basis, and one assumes, makes a modest profit - or at least plans to. There's performance, scalability, database infrastructure etc. Some of this you can mitigate by good design - caching, data normalisation etc, but some things you just have to pony up with the cash and pay to have it done properly.
Ultimately I started this effort because I was dissatisfied with the Suunto software but as the phrase in IT circles goes, "it works on my machine" and I need to consider both the time spent fixing 'public' issues and the costs associated.
Right now, this line of thought isn't really going anywhere, but I'd be interested in hearing thoughts on the commercialisation of such a product - either as a desktop application, or a web app (probably with some kind of sync agent). I appreciate that it's not all that easy when most of you get a poor UI experience at this stage, but mostly the graphs render and although you can't change the display period you should get some idea how it looks. The blog postings to date show some evolution too.
Random musings at the moment, probably because I'm annoyed with this UI problem, it's a cold, wet February and I work in an industry where large chucks are going down the toilet and dragging the rest of the economy with it.
Apart from that though, it's fine!
4 comments:
VPS Hosting, such as fsckvps, would probably be the way to go at first. Assuming the second option, that would mean recovering US$20/month in operational costs, in addition to your time.
I've had a bit of a think about this kind of thing from the perspective of an app for dive computers, but gave up more because of the data transfer issues than the web-app ones. From a data retention perspective, you could simply provide an export function for a start, then consider automated or semi-automated synching later. The two fears from an end-user's perspective come down to not being able to access their data and other people being able to access it.
You'll also need to consider the graph/chart options for web apps. I'm in the Java world and it's pretty limited unfortunately. Let me know if you want help or suggestions.
In the first instance, yes there are many suitable providers - I'm using discountAsp right now that have always proved solid in the past.
Data transfer can be solved pretty easily with local agent and some webservice stuff, but storage mounts up - my runtme data for the fat client runs to about 50Mb for around one year. Get 20 heavy users going and it mounts up quite quickly. I checked out fsckvps - good pricing, but site owner is responsible for backups etc - for static content that's not too much of a problem, but throw SQL transaction logs into the mix and it all gets a bit more complicated (and don't for get test recovery attempts either!)
As for fears, the authoritative source of the data would always have to be Stram, so that's not such an issue. Data segregation - yeah, but my day job is security risk so I know a thing or two about that (although it doesn't mean I'll get it right)!
Actually one of the interesting things about this is the type of anonymized statistics that you could gather across the pool of users - assuming permission of course. Are we all improving? What's our average TE's? How many miles are run with an HR between X&Y in the space of a week. I'm sure there are uses for such aggregated data that I can't think of at this stage.
Thanks for the comments
Personally, I prefer a thick client and local storage of data. Since I need access to my PC to upload the data in the first place, I prefer all the data stored there. In terms of data integrity, some sort of archiving utility that would allow me to backup data with a "single click" would be ideal. Not on topic, but what I'm looking for is not only better analysis, but also the ability to plan for training. Additions to create a training plan, and have it appear in the calendar would be great.
I'd be thrilled with either option. There are pros and cons to local data vs web app. Maybe we can find resources to help solve the ribbon problem...do you think you can describe it well enough for reporting? I have some good software dev't friends-- maybe they could help if you'd like another opinion on potential solutions.
What is your machine/operating system? Interesting that you don;t have the red-X. Can you give a generic explanation of what the code is to make the red-x-- then I could try to get some help.
Thanks for your efforts-- sorry for the frustration... if it's any consolation, at least a few of us really do appreciate your generous spirit!
David
David
Post a Comment