PolyCert: Polymorphic Self-Optimizing Replication for In-Memory Transactional Grids.
M. Couceiro, P. Romano and L. Rodrigues
Selected sections of this report were published in the Proceedings of
the ACM/IFIP/USENIX 12th International Middleware Conference, Lisboa,
Portugal, December 2011.
Abstract
In-memory NoSQL transactional data grids are emerging as an attractive
alternative to conventional relational distributed databases. In these
platforms, replication plays a role of paramount importance, as it
represents the key mechanism to ensure data durability. In this work
we focus on Atomic Broadcast (AB) based certification replication
schemes, which have recently emerged as much more scalable alternative
to classical replication protocols based on active replication or
atomic commit protocols. We first show that, among the proposed
AB-based certification protocols, no-one-fits-all solution exists that
achieve optimal performance when considering heterogeneous workloads
produced by complex transactional applications. Next, we present
PolyCert, a polymorphic certification protocol that allows for the
concurrent coexistence of different certification protocols and that
relies on machine-learning techniques to determine the optimal
certification scheme on a per transaction basis. We design and
evaluate two alternative oracles, based on parameter-free machine
learning techniques that rely both on off-line and on-line training
approaches. Our experimental results demonstrate the effectiveness of
the proposed approach, highlighting that PolyCert is capable of
achieving performance extremely close to those of an optimal
non-adaptive certification protocol in presence of non heterogeneous
workloads, and significantly outperform any non-adaptive protocol when
used with realistic, complex applications that generate heterogeneous
workloads.
Also available extended report (pdf)
Luís Rodrigues