quote:Since we’re brainstorming, here are some issues:
Originally posted by Jabbergah:
AS's ENTIRE customer base need not be contacted in such situations, only those with imminently pending snipes. The "bulk" e-mails could be batched by who has a pending snipe within the next 15 or 30 or 60 minutes (whichever increment results in a workable number for the bulk e-mailing). The next batch would be those who have a snipe scheduled in the next time increment, and etc. to some arbitrarily set advance time (maybe 4 - 6 hours out, or maybe this could be set when the routine is initiated). As each hour (or maybe time increment) passes without the problem being resolved, the next batch of customers with snipes scheduled 5-6 hours out would be e-mailed.
- Someone might place their snipe right before the auction ends. Wouldn’t the routine need to also monitor all new snipes as they are being placed so the user could be warned/emailed? If *everyone* received an email, then all new snipes wouldn’t need to be monitored, but then that’s back to the “bulk” email problem.
- What duration of downtime would warrant sending emails? 30 minutes? 1 hour? 2 hours?
- How do you determine, in advance, how long the service will be down?
- Do you want to send emails to users if the service is only going to be down for 15 or 30 minutes? Might not mass hysteria kick in? What if it turns out it’s only down for 5 minutes (allowing for the time it takes the programmers to make some coffee)?
- Of course, there are the few poor souls that can’t sign onto eBay at work and place a manual snipe or a proxy bid, so with the uncertainty of the duration of the down time, would it be best to email them even if their snipes are say 10 hours out (allowing them enough time to work, get home, read their emails and place a manual snipe or proxy bid)?
- Would you also send out a “The sniping system is back up” message?
- How would you respond to those that received a warning email and then placed a proxy bid which started a bidding war and were then notified that AS was back up within time for AS to have placed their snipe? There’s also the issue that AS would charge them for the snipe even if the proxy bid won.
- What happens if the problem is with the server, or servers, that stores the snipes? Put another way – how can the upcoming snipes be determined when the thing that knows about the upcoming snipes is down?
- Do you want the AS techie/techies using brain cells thinking about when/how/if to implement the notification system, or spending their full energies fixing the problem?
- What is the priority of this improvement/enhancement? Does it make sense to spend programming resources addressing this, or can the time be better spent elsewhere? (Infinite needs – finite resources)
I suspect that any system implemented will only replace the current ruffled feathers with other ruffled feathers (same bird, just different feathers).