iconimage - Fotolia

Cloudian CEO: AI, IoT drive demand for edge storage

Describing AI as an 'infinite data consumer,' Cloudian CEO Michael Tso said there's growing demand for edge storage because there's too much data being created to efficiently move.

AI and IoT is driving demand for edge storage as data is being created faster than it can be reasonably moved across clouds, object storage vendor Cloudian's CEO said.

Cloudian CEO
Michael Tso said "Cloud 2.0" is giving rise to the growing importance of edge storage among other storage trends. He said customers are getting smarter about how they use the cloud, and that's leading to growing demand for products that can support private and hybrid clouds. He also detects an increased demand for resiliency against ransomware attacks.

We spoke with Tso about these trends, including the Edgematrix subsidiary Cloudian launched in September 2019 that focuses on AI use cases at the edge. Tso said we can expect more demand for edge storage and spoke about an upcoming Cloudian product related to this. He also talked about how AI relates to object storage, and if Cloudian is preparing other Edgematrix-like spinoffs.

What do you think storage customers are most concerned with now?
Michael Tso: I think there is a lot, but I'll just concentrate on two things here. One is that they continue to just need lower-cost, easier to manage and highly scalable solutions. That's why people are shifting to cloud and looking at either public or hybrid/private.

Related to that point is I think we're seeing a Cloud 2.0, where a lot of companies now realize the public cloud is not the be-all, end-all and it's not going to solve all their problems. They look at a combination of cloud-native technologies and use the different tools available wisely.

Cloudian CEO Michael TsoMichael Tso

I think there's the broad brush of people needing scalable solutions and lower costs -- and that will probably always be there -- but the undertone is people getting smarter about private and hybrid.

Point number two is around data protection. We're now seeing more and more customers worried about ransomware. They're keeping backups for longer and longer and there is a strong need for write-once compliant storage. They want to be assured that any ransomware that is attacking the system cannot go back in time and mess up the data that was stored from before.

Cloudian actually invested very heavily in building write-once compliant technologies, primarily for financial and the military market because that was where we were seeing it first. Now it's become a feature that almost everyone we talked to that is doing data protection is asking for.

People are getting smarter about hybrid and multi-cloud, but what's the next big hurdle to implementing it?


Tso: I think as people are now thinking about a post-cloud world, one of the problems that large enterprises are coming up with is data migration. It's not easy to add another cloud when you're fully in one. I think if there's any kind of innovation in being able to off-load a lot of data between clouds, that will really free up that marketplace and allow it to be more efficient and fluid.

Right now, cloud is a bunch of silos. Whatever data people have stored in cloud one is kind of what they're stuck with, because it will take them a lot of money to move data out to cloud two, and it's going to take them years. So, they're kind of building strategies around that as opposed to really, truly being flexible in terms of where they keep data.

What are you seeing on the edge?

Tso: We're continuing to see more and more data being created at the edge, and more and more use cases of the data needing to be stored close to the edge because it's just too big to move. One classic use case is IoT. Sensors, cameras -- that sort of stuff. We already have a number of large customers in the area and we're continuing to grow in that area.

The edge can mean a lot of different things. Unfortunately, a lot of people are starting to hijack that word and make it mean whatever they want it to mean. But what we see is just more and more data popping up in all kinds of locations, with the need of having low-cost, scalable and hybrid-capable storage.

We're working on getting a ruggedized, easy-to-deploy cloud storage solution. What we learned from Edgematrix was that there's a lot of value to having a ruggedized edge AI device. But the unit we're working on is going to be more like a shipping container or a truck as opposed to a little box like with Edgematrix.

What customers would need a mobile cloud storage device like you just described?

Tso: There are two distinct use cases here. One is that you want a cloud on the go, meaning it is self-contained. It means if the rest of the infrastructure around you has been destroyed, or your internet connectivity has been destroyed, you are still able to do everything you could do with the cloud. The intention is a completely isolatable cloud.

In the military application, it's very straightforward. You always want to make sure that if the enemy is attacking your communication lines and shooting down satellites, wherever you are in the field, you need to have the same capability that you have during peak time.

But the civilian market, especially in global disaster, is another area that we are seeing demand. It's state and local governments asking for it. In the event of a major disaster, oftentimes for a period, they don't have any access to the internet. So the idea is to run in a cloud in a ruggedized unit that is completely stand-alone until connectivity is restored.

AI-focused Edgematrix started as a Cloudian idea. What does AI have to do with object storage?
Tso: AI is an infinite data consumer. Improvements on AI accuracy is a log scale -- it's an exponential scale in terms of the amount of data that you need for the additional improvements in accuracy. So, a lot of the reasons why people are accumulating all this data is to run their AI tools and run AI analysis. It's part of the reason why people are keeping all their data.

Being S3 object store compatible is a really big deal because that allows us to plug into all of the modern AI workloads. They're all built on top of cloud-native infrastructure, and what Cloudian provides is the ability to run those workloads wherever the data happens to be stored, and not have to move the data to another location.

Are you planning other Edgematrix-like spinoffs?
Tso: Not in the immediate future. We're extremely pleased with the way Edgematrix worked out, and we certainly are open to do more of this kind of spin off.

We're not a small company anymore, and one of the hardest things for startups in our growth stage is balancing creativity and innovation with growing the core business. We seem to have found a good sort of balance, but it's not something that we want to do in volume because it's a lot of work.

Dig Deeper on Storage architecture and strategy

Disaster Recovery
Data Backup
Data Center
Sustainability and ESG
Close