Deduplication
Cross-flow deduplication removes duplicate byte patterns from network traffic, whatever protocol, application or session originated the data. Patterns are saved on disk for future matches, and Wanos is designed to match the asymmetrical nature of network traffic.
Deduplication level 4 is the recommended setting: it applies to most scenarios and deduplicates at the highest level. Level 1 suits high-speed or limited-resource scenarios.
Stream compression
Real-time stream compression of network data gives higher compression ratios, acceleration and bandwidth savings. A higher compression level leads to a higher compression ratio; a lower level leads to higher throughput. Using the most efficient techniques available enables Wanos to run on low-cost ARM embedded systems.
We are impressed about the levels of compression. We did a test with the ntttcp Windows tool and we are amazed with the results: 80 Mbps got compressed over 30 Mbps.
Mijahil Miranda, Bolivia
Persistent data store
A large disk-based data store enables long-term retention of deduplication data and full service restoration after infrastructure outages. When the datastore is full it reaches 100% efficiency and manages the byte-cache first in, first out: new data replaces old data. Use solid-state drives where possible for best performance.
Results
At AIVA, replication approaches 40 Mbps while the WAN remains at 11 Mbps, resulting in 1.2 TB saved for the week. See the results.