Pythian provides invaluable Backup & Disaster Recovery (BDR) consulting for a company with no backup or recovery capabilities

Customer Success

Pythian provides invaluable backup & disaster recovery (BDR) consulting

When a software-as-a-service (SaaS) company’s HBase database became so massive it was unable to perform Hadoop cluster backups, they turned to Pythian to find a robust and cost-effective solution.


A large SaaS company’s on-prem Hadoop cluster was used to store tens of thousands of audio files generated by their proprietary software, and the associated metadata. However, the company’s 215-plus terabyte Apache HBase database was so large it was unable to support the necessary snapshots long enough to perform Hadoop cluster backups. If the Hadoop Distributed File System failed, they were facing catastrophic data loss and a recovery process of several weeks if not months.  

The company had previously attempted to fix the problem using Commvault, but there were too many cluster performance issues for this option to work. 

They needed a robust and cost-effective BDR solution. Pythian was there to provide unbiased advice. 

What we did

  • Presented several on-premises and cloud options 
  • Provided expert BDR advice to help the client navigate their options; ultimately, a cloud-native object storage/column store option was recommended

Technologies used

  • Cloudera 
  • Apache Hadoop 
  • AWS RedShift 
  • DynamoDB and Simple Storage Service (S3) 
  • Google Cloud BigTable 
  • BigQuery and Google Cloud Storage 
  • Commvault 
  • Swift 
  • Ceph 
  • MariaDB/AX 
  • ClickHouse 

Key Outcomes

The client’s IT leadership was able to verify their assumptions through Pythian’s vendor-agnostic expertise. This empowered the team to make confident decisions about their technology choices

12 hours or less to recover the last 12 months performance benchmark met

Explore our Database Services 

No matter your business, no matter the challenge: Pythian’s solutions drive results. 

More customer success stories