I think the time to charge a fee has arrived!

[tcgraham]

I am finding that SG performance continues to degrade. My concern is how much time time passes before a disposable address message gets back through your system. This is the senario:
1. I send a message using a disposable address to a website asking a question.
2. The website responds within seconds that they have sent me a response.
Before there response actually gets to my inbox 30-60 minutes might pass.
The problem is most prevelent with the above senario. Normally I don't need an immediate response.

There are lots of places a message can bog down.

But I just ran a test.
It is Sunday at 11:01EST. I sent a message to one of my disposable addresses at 10:54. It is now 11:10 and no response message. Granted only 15 minutes have passed but I expect 30 minutes will pass before I get a response.
SG is great and I have used it for years. But user growth may be causing a problem. What performance tests have your team conducted? On "average" how long would you expect a message turnaround to take? If you charged $5 a year to 140,000 users, you would have tons of money to get great performance. Or even just ask for a $5 donation. The donation approach would remove any accounting duties for you. Thanks for your thoughts. Tom - Florida

[josh]

The slowdown is not caused by the users (at least, not ~99.9999% of them) - most of the reason for delay is the huge botnet that's still launching dozens or hundreds of connection attempts at our server per second -- "legitimate" servers have to compete with this. We *have* managed to block a great deal of it before it gets into the code, which has resulted in much better (basically instantaneous) performance once a message gets through.

Charging users wouldn't improve this situation. Starting a new service with new domains and a new server and charging for accounts would likely provide great performance until such time as the botnet aims at the new server.

We're currently looking at hardware options that would increase cost, both up front and on a monthly basis, but show promise. I don't think the increase in cost would require charging users, though -- I think we could just do a brief fund drive.