A flexible framework for accurate simulation of cloud in-memory data stores

Abstract

In-memory (transactional) data stores, also referred to as data grids, are recognized as a first-class data management technology for cloud platforms, thanks to their ability to match the elasticity requirements imposed by the pay-as-you-go cost model. On the other hand, defining the well-suited amount of cache servers to be deployed, and the degree of in-memory replication of slices of data, in order to optimize reliability/availability and performance tradeoffs, is far from being a trivial task. Yet, it is an essential aspect of the provisioning process of cloud platforms, given that it has an impact on how well cloud resources are actually exploited. To cope with the issue of determining optimized configurations of cloud in-memory data stores, in this article we present a flexible simulation framework offering skeleton simulation models that can be easily specialized in order to capture the dynamics of diverse data grid systems, such as those related to the specific (distributed) protocol used to provide data consistency and/or transactional guarantees. Besides its flexibility, another peculiar aspect of the framework lies in that it integrates simulation and machine-learning (black-box) techniques, the latter being essentially 1 arXiv:1411.7910v1 [cs.PF] 28 Nov 2014 used to capture the dynamics of the data-exchange layer (e.g. the message passing layer) across the cache servers. This is a relevant aspect when considering that the actual data-transport/networking infrastructure on top of which the data grid is deployed might be unknown, hence being not feasible to be modeled via white-box (namely purely simulative) approaches. We also provide an extended experimental study aimed at validating instances of simulation models supported by our framework against execution dynamics of real data grid systems deployed on top of either private or public cloud infrastructures. Particularly, our validation test-bed has been based on an industrial-grade open-source data grid, namely Infinispan by JBoss/Red-Hat, and a de-facto standard benchmark for NoSQL platforms, namely YCSB by Yahoo. The validation study has been conducted by relying on both public and private cloud systems, scaling the underlying infrastructure up to 100 (resp. 140) Virtual Machines for the public (resp. private) cloud case.

An Evaluation of the Challenges of Multilingualism in Data Warehouse Development

“…Infinispan by JBoss/RedHat (Team, 2015) is a popular open source distributed in-memory key/value data store (Di Sanzo et al, 2014) which enables two ways to access the cluster: i) the first way enables an API avaliable in a Java library ; ii) the second way enables several protocols, such as HotRod, REST, Memcached and WebSockets, making Infinispan a language independent solution. Besides storage services, the middleware can execute tasks remotely and asynchronously, but end-users must implement Runnable or Callable interfaces.…”

“…The authors use Terracotta framework (Terracotta Inc., 2008) as a kernel of the entire solution. Gelibert et al (2011) point out limitations of using static types in the code, since instrumentation is done at runtime, thus the compiler cannot perform static verification on the application code. This creates complicated debugging scenarios when problems, especially transient ones, occur.…”

Proactive Cloud Management for Highly Heterogeneous Multi-cloud Infrastructures

“…In [16], [17] the authors propose to use a mixture of simulation and machine learning to study optimal deploys of cloudbased in memory applications. ACM Framework keeps the ability offered by these proposals and offers the possibility to modify the deploy at runtime in case the workload conditions change during the lifetime of the system.…”

Hybrid Machine Learning/Analytical Models for Performance Prediction

“…Normally, AM is exploited to capture performance of components whose internal dynamics are known and easy to monitor; ML, conversely, is typically employed to predict the behavior of components whose internals are hidden (and, thus, not observable), or whose performance dynamics are too complex to be accurately captured via analytical methods. This approach has been implemented for modeling performance of distributed transactional Cloud data platforms, where details of the underlying physical architecture is hidden by the virtualization layer; it has been integrated in hybrid models encompassing both queueing-theory based AMs [5,8] as well as simulation-based ones [3].…”

“…Gray box modeling [5,6,4,8,7,10,14,9,3] has emerged in the last years as an attempt to achieve the best of the AM and ML world. It relies on exploiting both methodologies, in order to compensate the weaknesses of the one with the strengths of the other.…”