Hammerspace – Unstructured Data Anywhere

HammerSpace announced the release of their Data-as-a-Service (DaaS) platform at Tech Field Day 17.  They bring a wealth of knowledge, experience and innovation to the world of unstructured data.  Their goal is to provide access to unstructured data everywhere, regardless of which data center or cloud the data is stored in.

hammerspace

HammerSpace believes that “you shouldn’t have to think about getting to your data” and they do this by decoupling the data itself from the underlying storage.  The magic of that decoupling happens in separating the metadata and the data services from each other.  Essentially, HammerSpace delivers a global namespace that provides access to your data, wherever it resides.  The metadata service helps HammerSpace to understand where the data itself should be placed, but the global namespace means that as a consumer, the placement doesn’t matter.  You just consume your data, and HammerSpace takes care of the intelligence required to make sure that your data abides by the rules that are set (compliance, locality, performance, etc.) within the product.  The basic construct of the HammerSpace service looks like this:

hammerspace2

There are three major components of the HammerSpace platform.  The inner layer shown is the data virtualization layer, which was actually very familiar to some of my fellow TFD delegates as it is essentially the foundation of a previous company from which HammerSpace has deep ties with, Primary Data.  You can find older Primary Data presentations on the Tech Field Day site to get a background on where the data virtualization layer got it’s start, but at a basic level this is the technology that allows for data to move between different types of underlying infrastructure, both on-prem and cloud.  For HammerSpace though, the data virtualization layer consists of really cool technology that doesn’t solve the whole problem.  That is where the meta-data-as-a-service layer and the hybrid cloud data management layers come in. 

A common theme among all of the TFD17 presentations was the prevalence of AI/ML.  HammerSpace is no exception, and the addition of the middle and outer layers shows just how much AI/ML comes into play with this technology.  Metadata intelligence is built by scanning the underlying data, and this intelligence is provided as a service by HammerSpace.  This allows for data to automatically be placed based on rules that are governed by the metadata itself.  In a crude way, it seems similar to VMware Storage Based Policy Management, but allowing for storage across a hybrid cloud ecosystem with AI/ML being used to help govern the policies.  

Similar to cloud-based products, HammerSpace is deployed as a pay-as-you-go service on a consumption model.  From an infrastructure standpoint, it is a hybrid cloud product, as it consists of an on-prem appliance and a service that can be spun up within the largest public cloud providers.  The architecture of the HammerSpace Geo-spanning Namespace is show in the diagram below:

hammerspace-arch

I am excited to see the evolution and adoption of HammerSpace.  They presented some very cool technology, even if at times the presentation got a little bit lost (at least on myself) in weeds on the data side of things.  It would also have been nice to see more real world demos and use of the HammerSpace interface itself from an administrative perspective.  As a data consumer, all of this works stealthily in the background, but someone still needs to stand the service up and get it to work seamlessly as a solution for a particular use case.  I’m curious as to what that experience is like, and how day-to-day administration plays a part beyond that.  I’d also like to see some of the white-boarding that took place during TFD17 end up in a something like a whitepaper just to get a more concise view of the technical deep dive. 

My fellow delegate, Chin-Fah Heoh, published another take on HammerSpace that gets a bit deeper into the meat of the subject.  Check out the HammerSpace product overview from TFD17 below.  Further HammerSpace presentations can be found within the Tech Field Day Youtube page:

Disclaimer: I was invited by Gestalt IT to participate as a delegate for this Tech Field Day presentation. While my travel and accommodations were paid for by Gestalt IT, I was in no way compensated for this blog post.  I was not obligated by anyone to post about any of the presentations that occurred during Tech Field Day.  Any blog posts, tweets or other content related to my time at Tech Field Day are my own views for the sole purpose of creating consumable content.

3 thoughts on “Hammerspace – Unstructured Data Anywhere

  1. This is no “stealth mode” product or company. Its the same primarydata product with no takers (read paying customers) with same architectural flaws . Diagrams are deliberately misleading with no mention of custom pnfs or gateway server. There won’t be real live demos because isht doesn’t work. How do I know ? I tested it.

    1. I was just reviewing my experience with their tech field day presentation. Regardless of their previous ventures, I’m certainly willing to let the market and wider IT community judge the success of Hammerspace as it evolves beyond what I saw earlier in the fall.

Leave a Reply

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