I sure hope they have to report the business units results every quarter
Then we will all know the farce of Sun's marketing claims.
Sun Microsystems' utility computing operation is being turned into a separate cloud business unit lead by Sun's chief sustainability officer Dave Douglas. The Reg has learned Douglas will run Network.com, which provides hosted applications, servers and storage charged on a per-use basis, and will report directly to Sun chief …
Here are the 3 simple reasons why Sun's network.com service is unpopular:
1) It's an upload -> run -> download batch job processor
2) It's just an upload -> run -> download batch job processor
3) It's really just an upload -> run -> download batch job processor
Here are the 3 simple reasons why people love to use Amazon's web services:
1) It's great value
2) It's running your own operating system images (nice easy Linux)
3) It's got great storage and queue services to orchestrate everything
Sun's working on Project Caroline www.projectcaroline.net but taking so much time. The upside for Caroline is that you'll be able to run it on your own grid or Sun's grid - depending on which setup you prefer. So if Sun's prices are too high, maybe somebody will run Caroline on Amazon's Web Services. But then Sun would lose (again).
I believe your comparison of Network.com's current service with AWS is comparing apples and oranges. As you say, NW.com is specifically for batch job processing - it isn't designed for, quoting Amazon, "Web Service" hosting.
The NW.com core product is the Sun Grid Compute Utility. It's a grid based utility for people needing to get compute work done. The billing is per-CPU hour, not per-instance, which reflects the different target audience who aren't looking for a minimum amount of load handling on a web site, but have a number of compute jobs whose run time is probably finite and fixed, unlike that of a web service.
I've had a play with Project Caroline - it's an exciting research project and shows off some of the Platform-as-a-Service features such as completely dynamic networking. Looking at the forums, there are already people trying to replicate the infrastructure based on the opened code, which is interesting.
I'm not sure that AWS is the perfect underlying platform for Project Caroline software, nor do I think it's a platform that everybody can move to. There are clearly much lower level networking and storage blocks that they are using to build up the featureset, which I believe would be difficult or not as efficient to replicate.
Another point is that Amazon's secrecy over AWS is losing them customers in some sectors where regulations or customers need to ensure the security of data (for instance). This was evident talking to people at the recent CloudCamp London event and isn't something that AWS is catering for well. For plenty of people, the loyalty to one company or another isn't just based on price!
There are other providers springing up (Flexiscale in the UK and other US based firms) who will hopefully provide their own unique features and different pricing models, perhaps plugging some of the holes.
Biting the hand that feeds IT © 1998–2019