Episode 3 of TalkCrunch is up. We invited in two key Amazon S3 team members, Adam Selipsky (VP of Product Management and Developer Relations for Amazon Web Services) and Dave Barth (Product Manager for S3), to talk about Amazon’s exciting new Grid Storage web service.
As I wrote previously, S3 provides a terrific opportunity for startups with great ideas for a storage user interface to avoid building a back end storage infrastructure. Amazon is offering extremely low pricing and a very dependable infrastructure. For some people, S3 will allow them to launch a service that they otherwise couldn’t have built.
Nik Cubrilovic, the CEO of startup Omnidrive, was co-host of the show (as was Keith Teare) and we tried to convince him, unsuccesfully, to switch to S3 for the back-end of his storage.
Talkcrunch continues to do well after its first three shows. It is currently included in the iTunes “new and notable” section and is on of the top 20 technology podcasts (as of right now it is #12).
I don’t think you could use S3 for a customer-facing hosted storage.
Costs aside:
1. Any data to/from the customer will make two round trips. Customer to Omnidrive (and its peers), Omnidrive to S3. Perhaps it would work as an archival-type nearline storage, or backup.
2. Amazon offers API overhead when all that Omnidrive needs is a raw block device or filesystem.
(I haven’t listened to the podcast yet, no speakers on this PC I’m borrowing…)
I think S3 is a great way for a startup to host media and have reasonable bandwidth costs. At $.20/GB xfer, that rings in much lower than most dedicated hosts.
It’s a good way to scale up until a company has enough clout to negotiate better bandwidth costs.
Initially I was excited about Amazon’s plans, and the price looked fantastic, especially the storage cost. I was even thinking the possibility of offering a 100 GB storage for $19.95/month.
But looking closer I discovered that the bandwidth cost is really not that great. The price of 20 cents per gig sounds enticing but don’t forget that there are three charges associated with bandwidth:
1/ inbound traffic to my server
2/ outbound traffic from my server
3/ inbound traffic to S3
[this is when traffic is going from customer to S3]
even if all three segments cost $.2/GB, we are looking at a total cost of $.6/GB. currently most hosting companies charge about $.5/GB for bandwidth. If someone can figure a way to send traffic directly to S3, then there might be a case.
D.
I’m still not sure of Amazon’s target market for this. The pricing is far too high for companies like Omnidrive to make money off of it.
I suppose it could be used as a small companies’ Akamei — for example a company wants to put out a largish video and doesn’t have the bandwidth to host it themselves. With the ability to make a bittorrent they could lower the costs of distributing it.
Amazon could try and offer a SQL API for the file storage so that a company could have a backup of their data (I say backup as the roundtripping involved would be pretty slow for real time use).
Amazon does have their brand name and ability to execute which allows them to charge more for their service. I doubt if the Omnidrives of the world can scale like Amazon does.
It’s too expensive, specially when you have to pay for both upload and downloads!!! Most hosting providers only charge you for downloads. With Amazon, you have to pay both ways. So the effective cost of adding 1TB of data for one month and then retreiving it is $150 + $200 + $200 = $550 which isn’t exactly cheap.
Amazon offers API overhead when all that Omnidrive needs is a raw block device or filesystem.
I suppose it could be used as a small companies’ Akamei — for example a company wants to put out a largish video and doesn’t have the bandwidth to host it themselves. With the ability to make a bittorrent they could lower the costs of distributing it.
I do not feel optimistic to it.
I suppose it could be used as a small companies’ Akamei — for example a company wants to put out a largish video and doesn’t have the bandwidth to host it themselves. With the ability to make a bittorrent they could lower the costs of distributing it.
I do not feel optimistic to it.