Reliable adaptable Network RAM

Abstract

Abstract-We present reliability solutions for adaptable Network RAM systems running on general-purpose clusters. Network RAM allows nodes with over-committed memory to swap pages over the network, storing them in the idle RAM of other nodes and avoiding swapping to slow, local disk. An adaptable Network RAM system adjusts the amount of RAM currently available for storing remotely swapped pages in response to changes in nodes' local RAM usage. It is important that Network RAM systems provide reliability for remotely swapped page data. Without reliability, a single node failure can result in failure of unrelated processes running on other nodes by losing their remotely swapped pages. Adaptable Network RAM systems pose extra difficulties in providing reliability because each node's capacity for storing remotely swapped pages changes over time, and because pages may move from node to node in response to these changes. Our novel dynamic RAID-based reliability solutions use idle RAM for storing page and reliability data, avoiding using slow disk for reliability. They are designed to work with the adaptive nature of our Network RAM system (Nswap), allowing page and reliability data to migrate from node to node and allowing pages to be added to or removed from different parity groups. Additionally, page recovery runs concurrently with cluster applications, so that cluster applications do not have to wait until all data from a failed node is recovered before resuming execution. We present results comparing Nswap to disk swapping for a set of benchmarks running on our gigabit cluster. Our results show that reliable Nswap is up to 32 times faster than swapping to disk, and that there is virtually no impact on the performance of applications as they run concurrently with page recovery.

Optimal Server Provisioning and Frequency Adjustment in Server Clusters

“…In [19], an open-loop control strategy was proposed to shift power between processor and memory to maintain a server power budget. In [20], researchers proposed a solution of storing page and reliability data to idle RAM to avoid using slow disk. A large portion of the power budget in server environments goes into the I/O subsystem, the disk array in particular.…”

Green Data Centers

“…The DPM scheme, being the only applicable mechanism in the storage domain [65], covers three levels cache, memory, and disk, which use hardware power management features, such as multiple power states, DRAMs [51], and multiple spin speed hard disks [66]. Moreover, RDRAM chip [67] (if used) in a memory system can be set to an appropriate power state independently. Thereby, enabling dynamic switching of RDRAM with power-aware page allocation in the operating system [68] becomes a feasible solution.…”

The Research on Cache Management of Cloud Storage

“…Similar to [14], [15] also provided remote memory as swap partition. It also employed the mechanism similar to parity logging to avoid the faulty of the swap-in node.…”