Our Technology

The Replify Accelerator performs two main tasks: reducing bandwidth usage; and improving end user experience.

Replify employs its bespoke Intelligent Caching Engine to achieve these, applying selective compression, appropriate identification and caching of application payload bytes and protocol specific enhancements to ensure efficient delivery or packets.

Protocol Optimization

Replify is able to identify a number of protocols and apply protocol-specific optimization. For example, the efficiency of the ‘chatty’ CIFS file sharing protocol is greatly improved by Replify’s read ahead and write behind optimizations. For HTTP, Replify is able to reduce the cost of messages by losslessly encoding lengthy header information. Many other protocols are optimized in a variety of specific ways but even without protocol specific optimisation, Replify still delivers world class WAN optimization for other traffic types.
Where the appropriate keys are provided, Replify is able to safely and securely accelerate SSL connections using the latest and most secure TLS protocols and features. In addition, Replify is able to secure normally insecure application traffic using TLS between Replify nodes.

Block Level De-duplication

The Replify Intelligent Caching Engine identifies which bytes are suitable for caching and efficiently hashes, stores and then transports the bytes to the connected Replify node.
The caching engine uses advanced rolling hash and indexing techniques to quickly and accurately locate blocks and transports small block references in place of the original data. Replify takes care of the integrity and synchronisation of the cache data, ensuring the correct data is always delivered quickly and efficiently to the client.
Unlike a traditional cache, the Replify engine works across protocols and for uploads and downloads. This means an file that is cached by Replify from an email attachment, will also be offloaded if the same file is upload or downloaded via HTTP or a file server.

Compression

We simply take the data emerging from the first and second optimisation stages above, and compress it using a lossless compression protocol. This typically gives a 50% reduction in data volumes for many common uncompressed data types such as XML, Office documents, database queries, file transfers and many web applications.

Erlang

Replify Accelerator is written in Erlang which is a language designed from the ground up for writing scalable, fault-tolerant , distributed, non-stop soft-realtime applications.
Erlang’s built-in concurrency support allows us to utilize the power of modern multicore systems with limited development effort. Combined with the OTP (Open Telecom platform) libraries that are distributed with Erlang, it is easy to rapidly develop multi-threaded applications that process network data at all levels easily and efficiently.
The Erlang VM has fantastic cross-platform support. It allows our optimization engine to be used on a variety of different OSes with minimal code changes required. The core engine is successfully deployed on Linux (various distributations), FreeBSD, Mac OS-X, Windows and Android.