TU BRAUNSCHWEIG
| Carl-Friedrich-Gauß-Fakultät | Informatik
Informatikzentrum

Replication-aware resource management

BetreuerVasily Sartakov
ProfessorProf. Dr. Rüdiger Kapitza
Projektnv-hypervisor
IBR GruppeDS (Prof. Kapitza)
ArtBachelorarbeit, Masterarbeit, Projektarbeit
Statusoffen

Overview

Non-volatile memory technologies are characterized by low endurance [1] and asymmetry in read/write latencies. The former feature makes high-availability (HA) techniques mandatory to use in persistent environments, while the latter makes impossible to use existing approaches developed for systems with symmetrical latencies. To overcome these issues we introduced multisite synchronous VM replication [3] -- a HA technique for persistent systems with asymmetric read/write latencies.

Multisite synchronous VM replication relies on modern network technology -- Remote Direct Memory Access (RDMA). Data between RDMA-enabled devices migrate in a zero-copy manner. Moreover, memory chunks can be linked together to perform a migration via a single transaction. Unfortunately, only 32 regions can be delivered together, and to increase replication performance, we merge some co-located blocks of dirty memory into bigger ones. However, since we transfer more data than need, the merging has an impact on an overhead of transferred data and replication performance.

One of the possible ways to increase replication performance -- reducing scattering of dirty pages, because co-location of dirty pages decreases overhead and increases replication performance. To make this, we need to make guest software 'replication-aware', e.g. by guiding a resource allocator (for example, memory allocator) with information which kind of object placement is more preferably from the replication point of view.

Under the umbrella of this concept, we offer several BA/MA projects aimed at the development of replication-aware VM software.



[1] S. Schechter, G. H. Loh, K. Strauss, and D. Burger. Use ecp, not ecc, for hard failures in resistive memories // in ACM SIGARCH Computer Architecture News, vol. 38, no. 3. ACM, 2010, pp. 141–152.

[2] M. K. Qureshi, M. M. Franceschini, A. Jagmohan, and L. A. Lastras. Preset: improving performance of phase change memories by exploiting asymmetry in write times // ACM SIGARCH Computer Architecture News, vol. 40, no. 3, pp. 380–391, 2012.

[3] V. A. Sartakov, R. Kapitza. Multi-site synchronous VM replication for persistent systems with asymmetric read/write latencies // The 22nd IEEE Pacific Rim International Symposium on Dependable Computing (PRDC 2017)




aktualisiert am 06.06.2018, 17:31 von Vasily Sartakov
printemailtop