Yeah, more RAM is linearly more costlier, which seems unfair and un-competitive. It does make me wonder though, how hard does Google (and others like linode) try to actually make all that RAM available to you. In other words, if you bought an instance with 4G RAM, do they absolutely guarantee you'll get all 4GB in physical RAM and won't start swapping because you're probably sharing the server with other apps?
On Wed, Dec 14, 2011 at 9:56 AM, bFlood <[email protected]> wrote: > exactly jeff, well put. > > On Dec 14, 10:36 am, Jeff Schnitzer <[email protected]> wrote: > > I think there is a legitimate gripe here which is that large-memory > > instances are unreasonably expensive. > > > > There's some significant value-add for GAE's "whole package" - > > automatic scaling, memcache, edge caching, deployment system, API > > access (although these APIs are generally charged separately). This > > makes the $60/mo for a basic (multithreaded) instance worthwhile. > > It's expensive but it's convenient, and most frontend work fits fine > > in the F1. Also it's a little bit of apples/oranges because the GAE # > > is heap whereas an Amazon # is VM size, but this is probably less than > > a factor of 2 difference. > > > > On the other hand, there are many application components whose primary > > requirement is a significant chunk of RAM. All that Google > > infrastructure is nice but it isn't nice enough to warrant a 10X > > premium just for a measly 1G of RAM. And you can't even get more. > > Seriously, a cheap amazon "standard" instance has significantly more > > RAM than the most expensive GAE instance... lame. > > > > Consequently, backends are useful as a long-running frontend, but > > absolutely useless as an in-memory index. We're priced into going the > > inconvenient route of placing memory indexes in other cloud services. > > > > I've been generally accepting of GAE's recent pricing changes, but the > > price of large-memory instances basically means I have to treat that > > option as if it doesn't exist. Which means when Google adds all these > > fancy features to support different kinds of instances, from my > > perspective, they're wasting their time. I can't use them until they > > make them cheaper. > > > > So here's my plea: a 256MB instance shouldn't cost twice as much as a > > 128MB instance, and a 512MB instance shouldn't cost twice as much as a > > 256MB instance. The price curve should drop off. There's a > > reasonable premium to pay for running on GAE, but a factor of 10 isn't > > it. > > > > Just for comparison... the largest GAE backend, at 1G, costs $460/mo. > > A 1.5G linode instance costs $60/mo. And I can get a 4G linode > > instance for $160/mo. And while it's not exactly an apples/apples > > comparison, when I need RAM, the priority of all those other Google > > niceties goes down considerably. And if I needed (say) four 1G > > backends, you can absolutely bet that I will go with Linode and pocket > > the extra $20k per year. > > > > Jeff > > -- > You received this message because you are subscribed to the Google Groups > "Google App Engine" group. > To post to this group, send email to [email protected]. > To unsubscribe from this group, send email to > [email protected]. > For more options, visit this group at > http://groups.google.com/group/google-appengine?hl=en. > > -- You received this message because you are subscribed to the Google Groups "Google App Engine" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en.
