"Derek J. Balling" <[email protected]> writes: > That happened to me when I tried to sign up, and it's the reason I ended up > with Linode. At the end of the day, when I need to crack open a new VM, 99% > of the time, I need it *now*, not in a couple weeks when the provider has > built out additional capacity.
Yup, I'm aware of how damaging it is to my brand and to the value I provide to customers. Now, I really do think that people should have more than one provider and be prepared to switch at any time, but that's simply not how most companies operate. It's a big problem; I can see the competition bearing down, I /will/ have to lower my prices again at some point in the next year or two, (that or get a business partner and start "adding value" rather than providing infrastructure... but that may as well be an entirely different business, one where I have little to contribute.) so now is the time to use revenue to buy more hardware. > I can respect the whole "we won't oversell" thing, a lot actually, but the > counterpoint to that is "you need to be ahead of your growth curve so you > don't leave potential-growth sales on the table, giving them to someone else." Well, really, with xen, you can't oversell ram. (you can oversell disk spindles and CPU, but I spend more on ram than disk and CPU combined, so you don't get as much mileage out of overselling those than out of overselling ram like you could with something like OpenVZ.) The problem is my incompetence, my innate conservatism, and my unwillingness to accept investment (and thus share ownership.) Really, if I can successfully overcome any one of those problems, I should be okay. I believe that if I quit screwing it up, I should be able to put up new servers as fast as people fill the old ones, you know, set up a virtuous cycle of increasing revenue. And my incompetence has been the problem there. It's almost to the point where I thought about buying dell. But, that would double my unit cost for servers, while figuring out the problems with a particular model is a one-time cost, so especially on what I'm paying myself now, it probably makes more sense to keep building, and just reduce the number of server models I have. As far as I can tell, I end up blowing another $250 or so for every 32GiB ram to use 32GiB servers rather than 64GiB servers, and I've already worked out all the problems with the 32GiB single-socket 8 core socket G34 systems, so I think I will just stick with those without letting myself get blocked by the problems with the dual-socket G34 systems. (this has been a mistake I made several times... when I switched to the G34 platform from the socket F platform that also held me up because rather than buying one G34 server and continuing to field socket F servers as needed until the G34 setup was solid, I bought several G34 servers, freezing all available working capital until I got the G34 platform stable enough to put customers on.) I'm learning, though, and things are getting better. We actually just brought a 32GiB server online, and filled it up from the waiting list without making a very big dent in the waiting list. we should have another up sometime next week (I've given up on the 64GiB box with twice the cpu/disk that was giving us so much trouble, and am moving the parts out to standard 32GiB servers that work with less trouble.) I'm working on simplifying the process of setting up new servers, reducing the number of hardware changes I make (standardization really does help a lot more than it seems like it ought) and I hope that as I figure that out I'll be able to buy the next server with the proceeds from the yearly signups on the last server. -- Luke S. Crawford http://prgmr.com/xen/ - Hosting for the technically adept http://nostarch.com/xen.htm - We don't assume you are stupid. _______________________________________________ Discuss mailing list [email protected] http://lopsa.org/cgi-bin/mailman/listinfo/discuss This list provided by the League of Professional System Administrators http://lopsa.org/
