We'll be safe enough once we make the jump to hyperspace… I mean Hyper-Converged!

Yes that was a quote from Star Wars, but with the latest announcement the day after Star Wars day (May 4th be with you) I thought it was a perfect fit.

Atlantis Computing announced availability of its new all-flash performance solution, touting it to have a lower cost point of 50-90% against other hybrid and hyper-converged appliances on the market to date. Atlantis have released this appliance with partnerships from SuperMicro, HP, Lenovo and Cisco from a hardware stack, whilst partnering with Citrix and VMware for their hypervisors. This helps avoid vendor lock in, but what about Microsoft? This is something that will be coming at a later date according to twitter posts from Atlantis employees.

Some readers may be asking what a Hyper-Converged system is. Hyper-Converged is the old all in one solution, whereby an appliance contains its own storage and compute to complete a specific task/workload.

In previous solutions, where I have personally looked into Hyper-Converged appliances, I have found that the scalability, density and price points don’t make sense, especially within a virtual desktop space. What will be interesting over the coming weeks will be other vendor announcements of their hyper-converged solution, or why Atlantis HyperScale is just a “hype”. Time will tell, but I have high hopes for this solution.

Couple of items for thought:

  • On the Atlantis HyperScale technical forum there has been hints to a density of 608 users on an Atlantis HyperScale CX-12 (4 Node) with 384GB RAM per node appliance with a resiliency of one node failure. These figures are great for user density but it doesn’t outline just yet the full picture due to missing information. Seth Knox at Atlantis outlines this to be coming shortly within reference architecture guides. From reviewing the datasheets the following was noted:
Maximum Number of Virtual Desktops (per 4 node appliance) Task Worker (20GB, 1vCPU, 1.5GB RAM LoginVSI Task Worker)

 

Knowledge Worker (50GB, 2vCPU, 2GB RAM LoginVSI Knowledge Worker)

 

Power User (100GB, 2vCPU, 4GB RAM LoginVSI Storage Workload)

 

Stateless VDI 450 / 650 325 / 500 150 / 250
Persistent VDI 400 / 650 325/500 150 / 250

From my experience of delivering VDI solutions, the real world density would be somewhere between a knowledge worker and power user, lowering that 608 user figure to a more realistic amount.

  • Within a server virtualisation space these appliances are looking to host between 170 – 365VMs per 4 node appliance depending on server workload resource requirements. Again the reference architecture should outline how realistic this is.
  • Atlantis are promoting this appliance as a 12TB and 24TB option (with the other options of memory and vendor on top). The 12TB and 24TB capacity is estimated capacity after an estimated 70% of deduplication and a single node failure. Based on USX and ILIO deduplication experience this is easily possible.

On a final note, why not give it a go and see if it can transform your virtual environments performance, increase your VDI experience whilst allowing for persistent VDI sessions, or implement it in front of your disk crippled database server, I am sure trials are available. At its current price point there is nothing else near it in the market today!

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.