Amazon's cloud to host Windows Server
By Scott M. Fulton, III | Published October 1, 2008, 1:02 PM
After tests are completed sometime this fall, the cloud provider announced this morning, customers will be able to deploy complete Windows Server-based machine images to Amazon's high-capacity computing cloud, eliminating hardware costs.
In what could be a waterspout moment, if you will, for cloud computing, Amazon Web Services (AWS) developer Jeff Barr announced this morning that his operation is currently hosting a private beta of hosted Microsoft Windows Server instances. Within the next three months, AWS customers will be able to deploy machine images with 32- or 64-bit versions of Windows Server -- including high-performance packages -- to Amazon's cloud, to be hosted remotely.
The move will enable small businesses to deploy conceivably major server applications, such as Unified Communications or large SQL Server databases, paying only by the month for storage consumed and bandwidth used, without any hardware purchases for servers whatsoever.
"You will be able to use Amazon EC2 to host highly scalable ASP.NET sites, high performance computing (HPC) clusters, media transcoders, SQL Server, and more. You can run Visual Studio (or another development environment) on your desktop and run the finished code in the Amazon cloud," Barr told customers in a post to AWS' blog.
Already AWS hosts a variety of flavors of Unix systems, including Solaris, and makes available a choice of three Linux standard distributions upon which customers can build a cloud machine image (AMI). Amazon is not saying whether it will sell licenses for Windows Server or whether they must be purchased separately, though that's certainly not out of the question. In fact, it may be necessary, since Microsoft's current licensing policy for Windows Server -- even after a recent policy change to account for shifting virtual machine instances among processors -- remains almost comically confusing.
Amazon's Elastic Computing Cloud (EC2) services are sold using a two-tier formula, the first of which involves instances -- images that are roughly equivalent to the leasing of a computer processor, except in the virtual space. You can adjust how powerful you want that processor to be. You're then charged by the number of PUT and GET requests your system receives per month. Those charges are added to the storage space you lease from Amazon's S3 service.
With charges broken down in granularity in terms of cents (use this calculator from Amazon to see for yourself), the monthly charge for running a high-intensity computing operation can still be maintained to under $1,000 per month. You're still administering the applications on your machine images using your own personnel; and with Windows Server, you could conceivably deploy System Center and run it from a remote desktop as though it were on-site. Still, this makes certain high-intensity applications that were once the exclusive purview of big universities and financial institutions, suddenly feasible for smaller schools and local banks and credit unions.
At the Interop show in New York last month, there were multiple prominent hints at the likelihood of a Windows Server hosting deal for Amazon, though sources at the time cautioned us it was all still speculative.
Amazon is inviting interested customers to sign up for notification as to when Windows Server hosting will be made publicly available.
I think this is a perfect fit for call center scaling and seasonal efforts.
http://afewtips.com
Score: 0
|This is pretty cool. It will be awesome when this becomes the competitive norm. No more physical co-locating which is damn pricey.
"waterspout moment" ???? Congrats on being the first:
http://www.google.com/se...ment%22&btnG=Search
Score: 0
|"You're then charged by the number of PUT and GET requests your system receives per month. "
Yeah that won't be rife with abuse or configuration issues.
How about Amazon focus on 99.000% uptime, let alone offering more services?
Score: 0
|I think they probably are close to 99% uptime.
Generally speaking, most web hosting providers guarantee between 99% and 99.9% uptime for their services. Although 99% may sound quite great, the number can be deceiving. The differences experienced by .9% for business owners and customers alike can be huge, greatly impacting the level of commerce and satisfaction.
99.99% Uptime: There are approximately 4.38 minutes of downtime per month or .073 hours per month, which is equal to 52.56 minutes per year or .876 hours per year
• 99.95% Uptime: Approximately 21.9 minutes per month or .365 hours per month, which is equal to 262.8 minutes per year or 4.38 hours per year
• 99.9% Uptime: Approximately 43.8 minutes per month or .73 hours per month, which is equal to 525.6 minutes per year or 8.76 hours per year
• 99.5% Uptime: Approximately 219 minutes per month or 3.65 hours per month, which is equal to 2628 minutes per year or 43.8 hours per year
• 99% Uptime: Approximately 438 minutes per month or 7.3 hours per month, which is equal to 5256 minutes per year or 87.6 hours per year.
Refrences:
http://www.mybestratedwe...-uptime-guarantees.html
Score: 0
|In a virtualized environment, that's not at all hard to do, specially when you can keep a hot image backup ready and waiting to go, and it only exists as a file when not in use.
And in the worst case, restoring a corrupted environment is as simple as reloading a VM file, rather than what most are familiar with in the hassle of reloading and configuring an entire server environment from scratch.
Score: 0
|Amazon was down 2+ hours in Feb and 8 hours in July. The impact was across the web, including NY Times, Reddit, and hundreds of smaller sites.
So yeah just above 99.8%.
Score: 0
|VM doesn't help you when the sites are unavailable. If you have dynamic content and it's pushed to two sites, or caching service like akamai on hold to publish data, then you should be fine, but many of these sites relied solely on S3 and paid the price.
Score: 0
|