James Thew - Fotolia

News Stay informed about the latest enterprise technology news and product updates.

Telefonica installs Caringo Swarm object storage

Global telecom company Telefonica is hoping to expand its digital services, and installed Caringo Swarm object storage to get the job done.

Global telecommunications firm Telefonica is using Caringo Swarm software as the foundation for public cloud services that may require hundreds of petabytes of storage.

The Madrid-based Telefonica has services in 24 countries, with most of its revenue coming from Spain and Brazil. It is looking to expand its digital services in those countries, as well as other parts of Europe and Central and South America.

To build up its storage to host services, Telefonica installed Swarm object storage on Dell servers. It also uses Force10 switches and SonicWall firewalls for networking and security.

Object storage manages objects in a flat address space, making it easy to adjust to petabyte scale. Each object is assigned a unique identifier, which allows a server to retrieve it without needing to know the physical location of the data. These characteristics make it a good fit for cloud storage.

Telefonica tested Swarm for a year before putting it into production last September, according to senior IaaS solution architect Fernando de la Iglesia. He said his team has 400 TB installed now with plans to expand as it adds services in more countries.

Luis Villarrubia Grande, Telefonica’s head of global cloud product development, said his team bought Swarm object storage to build Amazon S3-like services for its customers. File sync and share is among its infrastructure as a service (IaaS) and platform as a service (PaaS) offerings. "We needed a baseline storage service, in this case Caringo," he said.

Grande said Telefonica had been looking at object storage for several years, beginning with EMC Atmos. He said his team picked Swarm for its functionality, stability and maturity.

"We were looking for software-defined storage, and we were looking to introduce a cloud approach. We wanted something that could scale, is multi-tenant, fault tolerant and intelligent. We like that Caringo is hardware independent," he said.

"This is a product we can offer for the enterprise. There is no single point of failure, and you can have hundreds of petabytes [PB] in one node. That’s important for us."

He also pointed to Swarm's erasure coding and the ability to manage it from a central console, and support for S3’s API as selling points.

"Security and data protection are important," Grande said. "You can constantly replicate between nodes in different data centers. That will help us add nodes in different countries based on our requirements in those countries."

De la Iglesia said the initial deployment is in Spain, and Telefonica is adding two more nodes to increase capacity to more than 1 PB in that country. He expects rapid expansion soon, with one customer in Spain estimating it will require 73 PB and a potential customer in the U.K. that would require more than 50 PB.

"We are starting small," he said of the 400 TB initial install. "One benefit of Caringo is you can start small and grow in 100 TB steps."

Telefonica is also adding Swarm nodes in the Americas, according to de la Iglesia.

"The way we are consuming storage, we see great use for object storage in the near future," Grande said. "Object storage makes it easy for us to add storage services. You can’t predict what will happen in the next two or three years in our market."

Next Steps

What makes object storage a good choice over block or file storage?

Dig Deeper on Public cloud storage

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

Well, just about every object-based storage software provider runs on industry standard hardware, supports object replication and object erasure coding and can grow their clusters from hundreds of TBs to hundreds of PBs. Every object-based storage software provider also claims some level of compatibility with the AWS S3 API, which is the de facto standard for object storage and many can use multiple, geographically dispersed sites for object replication. File sync and share is an application use case for object storage and not an inherent characteristic of the technology. Caringo does claim not to use a POSIX file system on the HDDs in their storage server nodes and writes directly of the HDDs.
Yes, they all run on "Industry Standard" hardware, but the point I think they were trying to make is that Caringo is a pure software play, and you can run it on any hardware that you can buy. You are not required to use their OEM'd hardware, which is what most of the Object Storage vendors force you to do. You either have to buy their software on their own hardware, or their "certified" hardware. Caringo will run on the hardware you choose, which is far more flexible and does not lock you in to a single vendor. The other big gotcha in Object Storage is that most of the vendors can only scale up at the storage pool level, and you cannot mix drive sizes within nodes. That is less flexible and requires you to buy more hardware (refer back to point 1) Caringo allows you to mix heterogeneous disk types (SATA, SAS, SMR, SSD) and sizes (4, 5, 6 8, 10TB) all in a single node. So if you want to expand your cluster you can just lifecycle smaller drives with larger drives. You don't have to buy more nodes to expand. There are lots of nuances of functionality that distinguish the different Object Storage products from one another, people need to pay a lot of attention to the differences and understand the implications of those on your specific use cases and functional requirements. IMHO