When you’re building a healthclub, you have to have enough stairmasters and treadmills to accomodate peak periods, but not so many that that you make your club too big. Real estate and utility bills may swamp you if you don’t have enough members paying dues. If you have 1,000 members, you might only need 20 treadmills because it’s unlikely that all 1,000 members will show up at the same time looking to jog or speed walk. But you never REALLY know.
Of course it helps to have experience. Big clubs like Bally’s and Lifetiem Fitness likely have it down to a science. They track how many members they have, how often members come, how long they stay, and what equipment they use. I assume they use this data to make improvements (from time to time) so that the members don’t have to wait too long for any particular apparatus.
Depending on what time of year you start your healthclub, the data might trick you into buying the wrong amount of equipment. In January, your club is much more crowded as people try to make good on their New Year’s resolutions. In August, your club might be less crowded as members choose to pursue outdoor activities. Likewise, demand for your club services probably peak in the hours right after work lets out, and again early on Saturday morning. While these peaks and valleys might be frustrating, I would imagine that they’re very conistent in healthclub business.
All this is a LONG way of leading into my topic for today - building the infrastructure and managed hosting environment for mycolts.net.
When we first built and launched mycolts, we had to guess at the number of members we’d have. We had to guess how often they’d visit and how many photos they’d upload, etc. in order to design a hosting infrastructure to support our community. As we look to add more teams to the mix, we have to get a handle on real numbers so that we can charge other teams appropriately for the resources they consume.
Here’a a look at some of our early data:
I took a snapshot from the last 30 days. Looks like 71% of our registered users have been active at some point in the past 30 days, but only 3% were active yesterday. While that number is lower than I want, it would indicate that we shouldn’t build our infrastructure based on “registered” users as that would probably lead us to create a larger hosting environment than we actually need to support the true number of SIMULTANEOUS users we may see from day to day.
We’ll keep tracking and we hope to start seeing some patterns, but it’s tricky. We’re still in pre-season, so anything can happen and probably will. We know our colts.com traffic will double from where it is now when the season begins. I expect mycolts traffic will grow too, but who knows? I wish I could see into the future and know exactly what to expect, but…well, there’s no substitute for experience, right?
While we don’t know exactly what to expect, we have come up with some creative hosting opportunities / solutions…which I may share in a future post.
[...] out Coyle’s blog and his most recent post about tracking the number of active MyColts users. Besides blogging about the development of MyColts, Coyle also discusses social networking and [...]
Pat,
Have you guys considered utilizing Amazon’s S3 and EC2 services?
For those that don’t know, Amazon S3 is their pay as needed storage at extremely low cost. It continues to serve the startups well because you only as as much as you use and the pricing structure is some of the best in the industry.
EC2 is their computing service, whereby you can startup and shut down a virtual server depending on when you need the bandwidth and processing power. It is also pay as needed and it is extremely affordable allowing many startups to not having to worry about predicting hosting peaks and valley.
I personally have been implementing S3 and EC2 on my startup but have not had the chance to fully utilize it yet because it is not needed at this moment (my startup haven’t launched yet).
Here are some more info directly from Amazon:
S3: http://www.amazon.com/S3-AWS-home-page-Money/b/ref=sc_fe_l_2/104-5015261-4310332?ie=UTF8&node=16427261&no=342430011&me=A36L942TSJ2AJA
EC2:
http://www.amazon.com/S3-AWS-home-page-Money/b/ref=sc_fe_l_2/104-5015261-4310332?ie=UTF8&node=16427261&no=342430011&me=A36L942TSJ2AJA
- Jimmy
Sorry, wrong link above for EC2. Below is the correct one.
EC2
Thanks, Jimmy. I did speak with Amazon about its storage solution. Right now they don’t support Microsoft .net…but maybe someday. It did seem like a good option for storage.
And yes, we are also looking at the virtual server model…makes a lot of sense, especially for a somewhat seasonal site like ours. You can add or remove servers as demand ebbs and flows.
Don’t forget that its not football season just yet. Your percentage should go way up once the season starts and you should see peaks on Sundays and Mondays. If 71% of your users were active in the last 30 days, I think you can expect to see anywhere from 60-70% of your users active on peak days. You could also get really viral during football season and see your user rate go up faster then you could predict. To me, its always better to have excess servers then not enough and have the site crash the first week of football. This happens with cbssportsline.com. Luckily they are established or they would have a hell of a time bringing back customers if they weren’t.