Share this post

July 15, 2015

The Purest IT Infrastructure

Priyadarshi Prasad - Atlantis


It is a truism to say that this is the most exciting time to be in the IT infrastructure business. New ideas are popping up like mushrooms and the din they are generating would likely put the big bang to shame.  

However, if you are the IT architect at your company, charged with ensuring that your IT can support and help grow your business, some of these ideas mean might engender a few questions such as:

  • Should you stay with the traditional approach where compute and storage are treated as siloes?
  • Within storage, is the traditional hybrid design good enough or should you consider all-flash.
  • At an application level, how is the mix going to change between classic applications and newer applications (like cassandra, hadoop) that are putting new types of workloads on the IT infrastructure.

As an IT architect, if you step back, and think about what would be the most ideal (the purest) IT infrastructure as it relates to compute and storage, I submit that the following would show up in your consideration lists:


  1. Solid State as the preferred persistent storage media. And this begs the question if the architecture you are considering is optimized for solid-state. A topic in itself but the key areas to look for are if the architecture makes effective use of deduplication/compression technologies to drive effective cost of solid state down, and deliver consistent sub-ms latency even under failure. 
  2. Reliability, availability and failure handling being managed from SW, as against being managed in and having a dependency on specific hardware/appliance.
  3. Easy scale out - capacity and performance. Scale-out is truly the building block of purest IT infrastructure.
  4. Getting rid of siloes - storage, compute, separate management planes for each of them and so on was yesterday's way. Tomorrow's way is, ok I'll say it, hyper-convergence.

In marketing speak, the above could be summarized with two questions:


  • Solid state optimized vs. disk/hybrid optimized: Does the product have a built-in solid state optimized architecture to deliver flash performance at the cost of disk.
  • Converged vs. siloed: Does the product seamlessly enable server-Storage-Storage Networking convergence (hyper-converged), or is it still servile to the traditional siloed complexity.
If I look at the key players in the storage space, here is how they appear on the above questions:
In the above quadrant, if you are in the disk/hybrid-optimized siloed quadrant (most organizations are there), moving up to the top or to the right is at best an incremental enhancement. The real transformation in your datacenter comes when you move both up and to the right. More and more customers are realizing that every day and making the move to the purest IT infrastructure they can deploy today to secure their tomorrow.
With its pedigree in VDI based on memory optimizations, Atlantis understands data reduction (deduplication and compression) like few others. With Atlantis USX and Atlantis HyperScale, Atlantis has brought the same aggressive data reduction capability for a more general purpose use case, thereby delivering all-flash hyper-converged appliance that are actually cheaper than hybrid systems. While talk about dedup efficiency is cheap, Atlantis is the only company to back its claim with 70% data reduction guarantee (excluding thin provisioning savings).
Interested in discussing more? Why not join the conversation at @AtlantisSDS or ask me at @Priyadarshi_Pd.
Current rating: 4.5 (8 ratings)