Jul 4, 2017

Komprise to maximize the value of data tiering

Komprise (www.komprise.com), new data management player, has recently participated for the very first time to the IT Press Tour. The company, founded in 2014 by senior executives, Kumar K. Goswami (CEO), Krishna Subramanian (COO), and Mike Peercy (CTO), has raised a total of $18.2M for 2 VC rounds and we find 2 interesting persons as investors: Bill Moore, one of the father of ZFS and co-founder of DSSD, and Sanjay Mehrotra, co-founder and CEO of SanDisk and currently CEO of Micron.

The 3 Komprise founders
The mission of Komprise is pretty clear in order to offer a new generation data management solution powered by analytics to reduce data storage infrastructure costs. By data management, we mean here a data tiering solution dedicated to file storage that migrates inactive or cold data to secondary storage such on-premise or cloud-based object storage. We can also associate the idea to be a modern HSM. The company was recognized recently in the SDS 120 Club by Condor Consulting Group and in the Cool Vendor 2017 by Gartner.

We recently interviewed Krishna Subramanian and made this dialog available on StorageNewsletter here.

When the team has designed the product they wished to offer automation at every level, seamless integration and usage, limitless scalability and above all to be simple to deploy and use. Tiering is not new but how Komprise designs and builds its solution is pretty unique. It's of course all about metadata with a transparent data movement capability not based on stubs or other similar techniques. Also there is no need for specific or dedicated appliance, no storage agents as well is necessary and above all no impact on the data path. Nothing changes for the applications as all data appear to be local. One of the key approaches is to use symbolic links to maintain a reference of files on primary storage and make an association with the Komprise engine.


Komprise has chosen a distributed architecture based on Elastic data store across all Observers that constitute a grid. An observer is a Komprise VM, stateless by design, to deploy on some machines that analyze NFS mounts and SMB shares and exchange information with other peers and talk with the central Director for observers' state and configuration. You can loose the director but the environment is not stopped and continue to provide the data hierarchy service. The migration information is directly maintained as a distributed information with Komprise Access Address built on Elastic. It seems that the design chosen is pretty efficient and robust. A technical paper is available here.

To validate the market adoption, the company has selected a few vertical markets such genomics, finance, public sector, engineering, M&E and oil & gas. Komprise has announced a european hannel strategy with the signature of 3 resellers: B-Cloud in Italy, also a Cloudian partner, Itiso in Germany, also a Cloudian partner and interestingly, a Versity partner as well, and SymStor in UK. I invite you to check the presentation delivered during The IT Press Tour below.

Share:

0 commentaires: