by, Adam Bogobowicz, Sr. Director of Product Marketing, Parallels

 

If you missed all the fun at Parallels Summit in Vegas last week here is a quick recap. It was all about cloud storage. Pure software magic at its very best.

 

Just imagine a rack of servers converted with a software RAID device into a single storage cloud; with storage on all the nodes now available across the rack, available for any of the VMs or containers running on that rack, ready for any container of VM connected to that rack, or available as a pure store for any application or backup system connected to that rack.

 

Since my affinity is for virtualization, my personal favorite is the fact that software storage abstraction not only frees up all the dormant storage in the datacenter but that it enables High Availability as a simple outcome of that abstraction.

 

On the main stage at Caesars Palace, we demoed it using a server rack with 5 nodes. We ran a container streaming a video on node 2. Then we connected the video stream to the main projector and showed heart beats for all the nodes and the container right below.

 

All it took then to show High Availability was to parade unannounced in front of the audience on the main stage during the keynote and pull out both power cords from the node running a virtual container. At that point the management node, aka virtual RAID, rechecked availability of data on server  and confirmed that it was down and found the container with the video stream missing. It than used distributed data from remaining nodes to restart a container on a new node and 12th seconds later  re-started video streaming on the projector.

 

I know I should stop writing right here because it does not get much better than that, but here is a quick calc for all of you appreciating datacenter operations. If you are running mirror RAID on a single server you are probably averaging at least 4 service impacting incidents a year.  If it takes you 2 hours per incident to recover the service, your customers are experiencing 8 hours of downtime a year, or  about 2 “9’s” of availability on your service, leaving you with a bunch of unhappy customers, and you are a hoster who has lost business. 

 

Now if you run High Availability cloud storage on that same rack. You will still get same number of incidents, that is just life and stuff happens, but now you are going to recover service in 10-30sec. dropping your total yearly downtime to less than 5 minutes. Now you tell me how many 9s that is.