Speaker
C.D. Tiwari
(EMBL-EBI)
Description
- Accessing large amounts of data in the cloud poses several problems:
- Many bioinformatics applications require POSIX access, which does not scale well. Re-writing the application is not always an option.
- Data sitting in the cloud costs money, whether it’s being used or not.
- An ideal solution in many cases would be to provide federated data access to data stored on-premises, with caching in the cloud to reduce latency and use of network bandwidth.
- We have been using OneData to provide federated, secure, scalable access from the cloud to our on-premises storage. OneData offers tuneable caching, block-level access, and many other features that make it very attractive for our use-cases
- I present the tests we have done with OneData, including performance measurements from our first production use-case, distributed assembly of marine metagenomes in multiple clouds. We discuss our plans for rolling it out as a production service in the coming year.
Author
C.D. Tiwari
(EMBL-EBI)
