Virtualization Technology News and Information
VMblog Expert Interview: Ten Questions on Persistent Storage Containers with FalconStor's CTO and VP of Engineering, Mark Delsman


VMblog recently caught up with industry expert Mark Delsman, CTO and VP of Engineering at FalconStor, to gain a better understanding of persistent storage containers.

VMblog:  So it is true, was your team the first to develop a persistent storage container?

Mark Delsman:  Yes, it was. We were working to solve the issue of making archive data truly portable - so it could survive 50 or 100 years across many generations of the underlying hardware and operating system versions. To develop a persistent container, we had to create a new format to store data within the container named Portable Storage Format (PSF). Think of the PDF file format and how it is independent of the system and applications - this is what we are working to achieve for archival data storage. Each container has the metadata along with the compacted data that describes the original format - allowing it to be reconstructed when retrieval is required.

VMblog:  So many startups and engineering teams have grappled with storage for containers, how was your approach different?

Delsman:  We fundamentally started from a different viewpoint - a data-centric approach - where data is the primary reason for the container and execution space second. Instead of trying to allocate data space from a local resource - the container has the payload. We've started with high-value static data, but the concept can easily be extended.

VMblog:  Containers are virtualized at the application layer - what are the most significant benefits?

Delsman:  Containers are more compact than a VM because of the OS abstraction and are easily understood by virtually every Cloud and OS environment. This approach gives the highly portable benefit we were looking for in a 100-year archive. In addition - the execution ability gives us some exciting and unique benefits - like being able to test the data for consistency and integrity throughout its retention lifecycle.

VMblog:  Why do you believe this container-storage approach will stand the test of time?

Delsman:  Containers have established strong industry support in a short time - and releases incorporating containers span startups to increase company offerings. In addition, the significant Open Source community involvement led us to conclude that containers will be useful for generations. While there may be improvements and changes in containers in the future - the base configuration appealed to us as the best bet for longevity. We also chose object storage as the destination for the containers, with virtually infinite expansion, very high reliability, and data protection as the best choice for a long-term repository.

VMblog:  Did your team have a unique skill set to take a different approach to build persistent containers?

Delsman:  The team has been working in the backup and data retention field for years, and fundamentally was well-equipped to look at the persistence problem with a unique perspective. Starting with a goal like 100-year retention got our creative juices working!

VMblog:  How do you track containers in the Cloud and keep up with them?

Delsman:  While we maintain an index of containers that we have exported to object storage, we do not limit tracking to the index. Our containers have an embedded "World-Wide Name" (WWN) - similar to fibre channel devices or Mac addresses. This gives us the ability to search through an object storage account or rebuild our indexes after a data set is moved.

VMblog:  Since containers are runtime environments, can you have data and applications within the same container?

Delsman:  Sure - this is an exciting part of the innovation. Applications that operate on the data set - such as consistency checkers, search, retention locks, etc. can easily be included when the container is exported to the object store.

VMblog:  Historically data and storage systems were interrelated, and containers break that paradigm.  How portable are these containers?

Delsman:  Containers are highly portable - but access rights or ownership of the data is a big topic. While every container will be encrypted before export - there will be requirements as the technology grows to show chain-of-custody or limiting data to a particular geographic region. The regulations around data may prove to limit portability more than the actual data format itself.

VMblog:  Can this new approach meet the technology iterations that occur in the storage market and still deliver seamless data access?

Delsman:  Yes - we absolutely believe it can. Object storage can provide the enormous scalability and abstraction from the hardware, and the container also provides an abstraction from the hardware and operating system. This combination should ensure the long-term preservation of your data.

VMblog:  If storage containers take off like VM and App containers did, are you going to integrate with Kubernetes or another data management platform? 

Delsman:  Yes - that would make sense to integrate with Kubernetes. Initially, we want to be able to convince our customers about the safety and security of their data, but I see this as an area for expansion in future releases.

Published Tuesday, August 25, 2020 7:33 AM by David Marshall
Filed under: ,
There are no comments for this post.
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
<August 2020>