Swiss bank moves away from file with Cloudian object storage

Swiss financial services provider PostFinance has deployed 525TB of Cloudian object storage in a general move away from file access storage as well as to support archiving for Hadoop and Apache Spark big data workloads.

PostFinance provides financial services to more than three million customers, has around 3,500 staff and operates two datacentres.

Migrations towards object storage are taking place for Hadoop and Apache Spark big data archives, as well as a move away from file storage towards object storage.

Key reasons for this are scalability not found in file and block access storage, and that S3-based object storage brings ease-of-use for application developers, said senior systems engineer Andre Kocher.

PostFinance uses NetApp for file storage and Hitachi HDS storage for block. The latter is used for key databases and there are no plans to migrate data off this hardware.

However, there are plans to move 100 million files over to Cloudian object storage, and that move will mostly affect existing NetApp hardware at PostFinance. So, why the shift to object storage?

“We’re always looking for what’s going to be in the future,” said Kocher. “And object storage is one of the technologies we identified.”

A key driver for PostFinance is that it is very simple to use for application developers. “With object storage, we can give lots of responsibility to application developers. They can create their own buckets as they work. Before they would have to request file shares or LUNs in storage.

PostFinance has deployed 12 Cloudian HyperStore object storage nodes with a total of 525TB of capacity – inside are SAS drives, and connectivity uses 10Gbps Ethernet. Cloudian’s HyperFile provides access to file data.

Cloudian’s core product is object storage based on the Apache Cassandra open source distributed database. It can come as storage software to be deployed on commodity hardware, in cloud instances or in hardware appliance form. Its Hyperfile file access – which is Posix/Windows-compliant – can also be deployed on-premise and in the cloud to provide file access.

At PostFinance, migration of file data is an ongoing process, while big data archiving is expected to go into production in April or May.

Another key benefit for PostFinance is that S3 storage works well with containers, said Kocher. “It’s easier to have an S3 endpoint than to have to map to persistent storage.”

Is Kocher able to put any numbers on the benefits gained by the move to Cloudian? “No,” he said. “I can’t put anything into figures, but a big advantage is that the in-built data protection in Cloudian means we don’t need backup for the platform.”

PostFinance’s Cloudian deployment uses 8+2 erasure coding, which sees data split into six fragments with two sets of parity data.