CISCO HyperFlex, are they late?

By | March 1, 2016

CISCO has announced HyperFlex

The new CISCO HyperFlex solution will be added on to their current product offerings.

In 2016 the hyper-converged industries will be around a $1.9 billion dollars industry. Nutanix is slated to be their main competition within this industry.

The current traditional Integrated systems account for $7.2 billion dollars.

CISCO has been getting pressure from the channel to produce a hyper-converged solution. Current CISCO UCS customers will be able to integrate HyperFlex directly into their current UCS environment. This will allow current customers to easily scale their infrastructure.  HyperFlex designed to support VMware, Hyper-V, KVM, and containers.  Currently only VMware is available.  There are plans to add Hyper-V and containers in the very new future.

HyperFlex-block

CISCO biggest play on this market is the claim they are a complete hyper-converged solution. This includes Software Defined compute, storage, and networking.  CISCO will be utilizing Springpath software within HyperFlex.  CISCO has re-branded Spingpath into CISCO HX Data Platform. There are no comments coming from CISCO that Springpath will be purchased in the near future.

The HyperFlex Foundation is built on:

  • Log-structured File System
  • Designed for distributed, scale-out storage
  • Inline data optimization impact
  • superior flash endurance

These features sound very familiar to other competitors in the hyper-converged space.

The HyperFlex solution is similar to other hyper-converged solutions. There is a three node minimum requirement. In order to Scale Cache or Capacity you will be required to add/purchase another node.

3Node

The HX Data Platform will stripes data across all nodes simultaneously leveraging cache across all SSDs for fast write speeds. This sounds like it could end up being a major bottleneck with storage as you add more nodes. CISCO also claims that there will be no performance impact with inline duplication or inline compression. I will have to see these features in production before I believe there is no performance impact.

The UCS manager and UCS director (coming soon) will be utilized to configure the HyperFlex solution. Once configured, HyperFlex has a vCenter plug-in that allows you to administer the full HyperFlex solution from one pane of glass. vCenter.

 

CISCO mentioned that clones and snapshots are pointer-based and space-efficient.  I have an issue with snapshots and backups. Snapshots are only from the point of time they were taken and forward from there. If the main files are lost, these snapshots are useless. I look forward to seeing this feature in production to see if it is more than a snapshot solution.

Ok I have one rant. CISCO states that you can plug HyperFlex into any network and it will work.  In an enterprise environment that has $30-90K invested in CISCO ACI solution, yes it will work. In very basic networks, small to medium-sized businesses, you find ghosts in the network configurations.  Things like routing issues, external IPs used for internal address, 22 bit subnets that were setup like 24 bit subnets.  These are issues that are outside the control/scope of any vendor.  Software defined networking cannot solve this issue.

I am glad that CISCO finally made it to the party. We will see if they are fashionably late or they missed the bus.