New ceo expects caringo swarm to ride object storage wave data recovery recuva

Barbagallo: We’ve been around for a long time. Object storage as back-end storage for cloud service providers — not unlike AWS and Azure — has been around for quite some time. But only in the last two or three years are we seeing object storage as an on-prem alternative to filers for unstructured data really taking off. My responsibility is to take advantage of this wave with a mature and well-rounded product.

In the last couple of years, we’ve really concentrated on taking a hard look at specific vertical segments, rather than trying to be a generic object storage for any kind of use case or workflow. We’ve been focusing on media and entertainment, as well as high-performance computing, and then the federal government space, which is a combination of both of those other two verticals around video surveillance and analytics.


Barbagallo: There’s really been no change in our partnership from a reseller standpoint. Their sales teams still resell us. If you’re out in the field and have a good relationship with a partner’s field personnel, and you know it’s worked for you in the past, you tend to go with what you know and what works. If it isn’t broken, why fix it?

But we’ve also started to forge partnerships with other resellers of our software, and we’ve also forged technology partnerships. That’s what we’ve been focusing on in the last couple of years, especially partners in the media and entertainment space — front-end providers of media asset managers and back-end repositories for those kinds of products.

Barbagallo: A little over 50. We’re still small and lean. We have emphasized working with partners since our inception. Dell was one of those partners, with 4,000 salespeople worldwide. Our goal was to rebuild our channel with multiple organizations, rather than just Dell.

We were profitable before the change of direction with Dell in 2015. We’ve become not profitable, but we expect to become profitable again by the end of the year. We’re now starting to see the fruits of some of our new partnerships. When you build out new partnerships, things take time to come to fruition.

Barbagallo: Anything that makes sense to store in a file folder hierarchy is most likely better served by network-attached storage. Anything in a database is more likely better served by a NAS or block-based storage area network. But we’re seeing sophistication of customers with more unstructured consumer data. There’s really not a big advantage to storing those reams of unstructured data in expensive NAS- or SAN-based filers. That’s what we’re starting to see.

Specifically in media and entertainment, with video at 4K and now going to 8K video resolution and all of these streaming services springing up, a NAS or SAN is overkill. In some cases, it’s even a detriment for storing large-scale and large volumes of data. One of the beauties of our object storage is, because it’s HTTP-based, we’re already essentially a content delivery network. We wipe out a whole layer of architecture having to install application servers in front of databases or servers.

Barbagallo: We do that primarily to bridge the gap of data ingest and egress to applications that need to act on that unstructured data. We’re working with a couple of customers where data typically comes in through more legacy environments. That’s where the requirement for NFS and SMB protocols lies.

Then, once in the object store, they want to run analytics — either Hadoop or they’re building custom analytics engines that are using the S3 protocol to access that data. It’s all nice and encapsulated over the network via HTTP. That’s looking to be the wave of the future, but there are so many applications and legacy hardware that need to send data to the data store, and it’s typically with a protocol they’re familiar with, like NFS or SMB.

We’re not trying to replace network-attached storage for people’s home directories. If any object storage vendor tells you they’re able to do that, they’re lying, because object storage is all about optimizing throughput by a number of clients interacting with storage at the same time. It’s not about clients interacting with a single file or collaborating with a single file. That’s where block-based storage has to shine.

Barbagallo: The advantage we have is we’ve been around a long time. Storage is hard. There are a lot of new vendors springing up, because object is becoming more popular and more mainstream, but they’re taking shortcuts. They’re writing an object storage app on top of a Linux file system. The whole point of object storage is to break the chains of file-system limits and the inode limits on a particular server.

We were one of the first to do that. We built our own file system based on key-value pairs. That takes a while to make sure it works on production capacity. That level of maturity works for us. We can improve by adding to the user experience. We can add support for controlling multiple clusters from a single user interface, as an example. We’re developing additional analytics or partnering with analytics companies, as machine learning and IoT become more prevalent.

banner