Not so sure, there... You pay for more than *just* bandwdth -- you pay for CPU-hours and each (million) of your IOs, too. I know of quite a few databases that can go through a million IOs in a real hurry. This is something that *could* be very interesting for low-volume applications. Training sand-boxes come to mind. Development systems. Maybe even test servers if they are used only infrequently. I do not profess to know much of anything about Amazon's EC2 offering, but having run a few numbers through their online pricing tool, it seems to me that EC2 will have a tough time competing with the purchase price of an 8-core Sun XFire 4x40 server, at least for a consistently busy database that you plan on keeping for more than a few months. On Tue, Sep 23, 2008 at 8:01 AM, Mark W. Farnham <mwf@xxxxxxxx> wrote: > seems to me the big catch is upload bandwidth. Now if they provide a way > to send in your initial bulk by sneakernet (or trucknet) so you can keep > your upload bandwidth down to something just proportional to changes, I > think this will indeed be the advent of a watershed event. Data will nearly > never be lost, and eventually everyone will have access to everything in the > cloud, by hook or by crook. > > > ------------------------------ > > *From:* oracle-l-bounce@xxxxxxxxxxxxx [mailto: > oracle-l-bounce@xxxxxxxxxxxxx] *On Behalf Of *Stephens, Chris > *Sent:* Tuesday, September 23, 2008 9:50 AM > *To:* oracle-l@xxxxxxxxxxxxx > *Subject:* wow. cloud computing is getting serious (oracle + amazon + > openworld announcement) > > > > http://aws.typepad.com/aws/2008/09/hello-oracle.html > > > > I guess the 'big' announcement isn't until tomorrow. > > > > For the first time in a while I really feel like we are seeing a new > paradigm in computing actually taking hold. …maybe I just don't have enough > caffeine in me. > > > > Chris > > > ... > -- Cheers, -- Mark Brinsmead Senior DBA, The Pythian Group http://www.pythian.com/blogs