I guess google $0.10 per hour instances are same as Amazon $0.10 instances.
The googles trick is that you do not pay for CPU but for request time. So you pay for IO wait time as well. When servlet waits for DataStore, then you still pay. On Ec2 you can have 10-50 requests concurrently waiting for DB at same time for same money. DB may costs a lot also. 5 second request may include 30 seconds of DB time. (e.g. 3x500 entity inserts in one request, bulk upload) Anyway app engine is great solution. It is often cheap or even free and reliable. Even after last outrage I can say it is reliable as far as far data is not lost. ---- I am not talking about production deployment right now. I think m1 may be able to do that job. With appengine I have to pay for bandwidth also. So I ignore it right now. On Mar 7, 4:58 am, Tracy Hurley <[email protected]> wrote: > I'm pretty sure you would need a 64-bit instance to do a database that > size. That requires doing at least the m1.large instance size, which runs > at about $8/day. To that, you would have to add the bandwidth for inserts > and deletes unless the machines doing them were running on the local > network. You also would probably want to use elastic block store (EBS) to > store the database since any data saved could be lost if the instance goes > away. EBS is charged at $.10 per GB-Month and $.10 per million I/O > requests. Amazon projects that a medium website database with 100 GB in > size and an average 100 I/Os per second would cost $10/month in storage > costs and $26/month for the I/O requests. Even if Mongo has similar usage > (I haven't tested that part yet), it could still be significantly better > than the Google appengine numbers although not quite the 5-10 times > mentioned. > > That said, it's completely possible to run Mongo on ec2 through Scalr even > if Scalr decides to not support it. It would be great if they did though. > One of the biggest problems is that high availability Mongo doesn't quite > seem to be ready yet. Replica-pairs are great, especially if you have > enough elastic ips available to tie one to each of the two instances. > > > > On Sat, Mar 6, 2010 at 6:06 PM, Max <[email protected]> wrote: > > It will add a great value to scalr if you will add mongo support. > > Mongo is fast and often is better then mysql. > > > Next step after mongo could be apache cassandra on scalr. > > > We are coding much on google appengine now and often appengine is > > expensive. > > Scalr could beat appengine costs 5-10 times. > > > e.g. > > DB size 30GB > > insert: 10GB per day > > delete: 10GB per day > > It costs $30-50 per day on appengine. > > > single m1 EC instance could do that as well. > > > or 2 m1 for replications > > > NoSql is great in many cases. > > > We have 10 GB table and it is imposible on mysql to add column without > > stopping service for long time. > > > Cheers > > > -- > > You received this message because you are subscribed to the Google Groups > > "scalr-discuss" group. > > To post to this group, send email to [email protected]. > > To unsubscribe from this group, send email to > > [email protected]<scalr-discuss%2bunsubscr...@goog > > legroups.com> > > . > > For more options, visit this group at > >http://groups.google.com/group/scalr-discuss?hl=en. -- You received this message because you are subscribed to the Google Groups "scalr-discuss" 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/scalr-discuss?hl=en.
