A second version of the Zswap patch-set for the Linux kernel was published this week. The Zswap patches provide compressed swap caching support to compress pages in the process of being swapped and compresses them into a dynamically allocated RAM-backed memory pool.
Zswap attempts to avoid writing back to the swap device in order to reduce I/O and yielding greater performance in scenarios where swapping does take place. When building the Linux kernel under heavy memory pressure, it's been reported by the lead Zswap developer that using the new technology can lead to a runtime reduction of 53% and an I/O reduction by nearly 80%.
The lead use-cases for Linux Zswap are systems with limited amounts of system memory, virtualization where VM guests are sharing common I/O resources, and systems backed by solid-state drives where reducing the number of writes to the disk can increase the SSD's life-span.
In Zswap v2, the set of nine patches have been re-based to the latest Linux "-next" kernel and there's various technical fixes to the code.
More information on the latest state of Zswap is available from this kernel mailing list post
for the newest patches. Zswap is a possible candidate for merging into the Linux 3.9 kernel.