AWS adds 32-vCPU option and an easier on-ramp to its cloudy desktops
- Reference: 1737095229
- News link: https://www.theregister.co.uk/2025/01/17/aws_workspaces_upgrades/
- Source link:
The cloud giant’s desktop-as-a-service offering is called “WorkSpaces” and on Wednesday two new instance types were [1]annoucned : the GeneralPurpose.4xlarge with 16vCPUs and 64GB of memory, and the GeneralPurpose.8xlarge which packs 32vCPUs and 128 GB RAM. Both also include a 175GB root volume – the virtual disk that includes the OS image – and 100GB of storage for users’ files.
The instance types are billed as suitable for the kind of applications that are often run on workstation-class PCs: engineering data-analysis tool Matlab, the statistical analysis package R, and big software compilation jobs all get a mention as suitable workloads.
[2]
The GeneralPurpose.8xlarge instance type is Amazon’s first to offer 32 vCPUs.
[3]
[4]
It’s not cheap when Windows is included, at $590 a month, and $295 for the 16 vCPU version. Hourly prices of $4.56 and $2.28 an hour are available after paying $19/month. Prices are lower if you bring your own Windows license – but by less than ten percent. This appears to be an Windows-only offer for now, as no pricing for the instance types is listed when running the Linux distributions supported on other WorkSpaces.
[5]AWS plays with Fire TV Cube, turns it into a thin client for cloudy desktops
[6]Omnissa, VMware's old end-user biz, emerges with promise of 'AI-infused autonomous workspace'
[7]Microsoft brings its cloudy virtual desktops on-prem to AzureStack HCI
[8]Streaming apps – and maybe even Cloud PCs – coming to electric cars
The two new instance types join the Power and PowerPro offerings – which include dedicated video memory – as earning Amazon’s recommendation for those who want “the best experience with video conferencing”.
Cloudy desktops can offer a speedy user experience, but can sometimes struggle with apps like video conferencing that need uninterrupted realtime data flows. Microsoft [9]admits its Teams collaborationware can’t deliver “optimized” video and audio chat on cloudy desktops, and also offers a [10]version of the software tailored for cloud PCs. The company suggests redirecting video and voice calls to the local device used to access cloudy desktops.
With its advice that only its mightiest WorkSpaces deliver “the best” vidchats, Amazon’s approach seems to be just to throw more compute resources at the problem.
[11]
Amazon’s not all about brute force with its cloudy desktops: Earlier this week it [12]announced tweaks to its Amazon EC2 Image Builder that allow direct conversion of Microsoft Windows ISO files to Amazon Machine Images, the virtual appliance format used to boot a VM in AWS.
Orgs with big Windows desktop fleets will generally maintain a “golden” image of their preferred PC setup. That can now be converted into an AMI used to boot an AWS Workspace, an arrangement that Amazon says will make it easier to bring your own license. It will also mean more consistency for cloudy PCs, meaning users' existing PC management tools and processes should apply - perhaps making cloud PCs more palatable? ®
Get our [13]Tech Resources
[1] https://aws.amazon.com/about-aws/whats-new/2025/01/larger-general-purpose-bundles-workspaces-personal-core/
[2] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/virtualization&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2Z4o4XXKFsntpXb-3spzv-AAAAMo&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/virtualization&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44Z4o4XXKFsntpXb-3spzv-AAAAMo&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/virtualization&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33Z4o4XXKFsntpXb-3spzv-AAAAMo&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[5] https://www.theregister.com/2023/11/27/aws_workspaces_thin_client/
[6] https://www.theregister.com/2024/07/26/omnissa_strategy_reveal/
[7] https://www.theregister.com/2024/02/06/azure_virtual_desktops_azurestack_hci/
[8] https://www.theregister.com/2023/05/02/streaming_apps_cloud_pcs_evs/
[9] https://learn.microsoft.com/en-us/azure/virtual-desktop/teams-on-avd
[10] https://learn.microsoft.com/en-us/windows-365/enterprise/teams-on-cloud-pc
[11] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/virtualization&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44Z4o4XXKFsntpXb-3spzv-AAAAMo&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[12] https://aws.amazon.com/about-aws/whats-new/2025/01/ec2-image-builder-converting-windows-iso-files-amis/
[13] https://whitepapers.theregister.com/
Re: How is this Cost-Effective for Businesses?
Rental is always expensive compared to buying, but there wouldn't be a business in it otherwise.
Though it does demonstrate a lot of the confused thinking around cloud - the model works great for things you only need temporarily, but if it's for anything truly permanent the direct cost benefit isn't exactly obvious.
Guess it's like any other rental or lease then.
It's not like the numbers are difficult to understand but there does seem to be a blindness to cost when it comes to paying for cloud services vs buying actual kit. Like it's not real money, at least up to the point where the massive accidental bills appear. Though again that's exactly how many people get into PCP/lease/phone contracts on a personal basis because they can't understand what the 'affordable' monthly bill really costs overall.
You use cloud for flexibility and the possibilities that gives, not because it's a cheaper way to build the same capacity as your fixed systems. As I repeatedly have to point out when explaining about my private cloud infrastructure when discussing costs.
You'd have to be stupid to pay that much to use someone else's computer instead of self hosting
For that kind of money, you could get a set up a GNUbooted KGPE-D16 with dual 6282 SE's to get 32 cores and install 128GB (or 256GB) of ECC-DDR3 and have control over the business's computing.
Even assuming those are half as fast as you need and you need 2 and the internet and power is expensive, that still works out to be much less expensive after a handful of months.
Even if the business doesn't want to have control over its own computing, you'll be able to find a second hand, or even brand new server that is better for cheaper (many businesses even have a perfectly usable server that is sitting there unused as windows eventually slowed down on it, but that turns out to be plenty fast once you install GNU/Linux).
Remote VMs only appear to ever make financial sense if you want to test something out for a few days or weeks before purchasing and installing hardware.
If you are wanting a 'workstation grade' computer, most people are wanting a dedicated GPU to go with their high spec CPU and since the article only mentions 'dedicated video memory' and not a dedicated GPU it sound like its not really a service that most people could use to replace their workstations.
And if your just wanting a Windows PC in the cloud to run Office 365 or similar on then its way overkill for that and cheaper alternatives are available. So i'm struggling to see who is actually going to want to use this?
How is this Cost-Effective for Businesses?
A top-of-the-line, non-AI-featured, gaming-optimized Intel Core i9-14900K retails for about US$450.
In three or four months, perhaps five, the fees for using an AWS GeneralPurpose.8xlarge VM would pay for a similar desktop system (Yes, AWS' offering is for 32 cores, vs the Core i9-14900K's 24 cores. But the AWS cores are virtual cores, with the performance reduction that implies.)
Or, is this offering aimed at businesses crippled by the OPEX/CAPEX artificial distinction?