Studying Recommender Systems to Enhance Distributed Computing Schedulers

Distributed Computing frameworks belong to a class of programming models that allow developers to

launch workloads on large clusters of machines. Due to the dramatic increase in the volume of

data gathered by ubiquitous computing devices, data analytic workloads have become a common

case among distributed computing applications, making Data Science an entire field of

Computer Science. We argue that Data Scientist's concern lays in three main components: a dataset,

a sequence of operations they wish to apply on this dataset, and some constraint they may have

related to their work (performances, QoS, budget, etc). However, it is actually extremely

difficult, without domain expertise, to perform data science. One need to select the right amount

and type of resources, pick up a framework, and configure it. Also, users are often running their

application in shared environments, ruled by schedulers expecting them to specify precisely their resource

needs. Inherent to the distributed and concurrent nature of the cited frameworks, monitoring and

profiling are hard, high dimensional problems that block users from making the right

configuration choices and determining the right amount of resources they need. Paradoxically, the

system is gathering a large amount of monitoring data at runtime, which remains unused.

In the ideal abstraction we envision for data scientists, the system is adaptive, able to exploit

monitoring data to learn about workloads, and process user requests into a tailored execution

context. In this work, we study different techniques that have been used to make steps toward

such system awareness, and explore a new way to do so by implementing machine learning

techniques to recommend a specific subset of system configurations for Apache Spark applications.

Furthermore, we present an in depth study of Apache Spark executors configuration, which highlight

the complexity in choosing the best one for a given workload.

Core and Shell Song Systems Unique to the Parrot Brain.

The ability to imitate complex sounds is rare, and among birds has been found only in parrots, songbirds, and hummingbirds. Parrots exhibit the most advanced vocal mimicry among non-human animals. A few studies have noted differences in connectivity, brain position and shape in the vocal learning systems of parrots relative to songbirds and hummingbirds. However, only one parrot species, the budgerigar, has been examined and no differences in the presence of song system structures were found with other avian vocal learners. Motivated by questions of whether there are important differences in the vocal systems of parrots relative to other vocal learners, we used specialized constitutive gene expression, singing-driven gene expression, and neural connectivity tracing experiments to further characterize the song system of budgerigars and/or other parrots. We found that the parrot brain uniquely contains a song system within a song system. The parrot "core" song system is similar to the song systems of songbirds and hummingbirds, whereas the "shell" song system is unique to parrots. The core with only rudimentary shell regions were found in the New Zealand kea, representing one of the only living species at a basal divergence with all other parrots, implying that parrots evolved vocal learning systems at least 29 million years ago. Relative size differences in the core and shell regions occur among species, which we suggest could be related to species differences in vocal and cognitive abilities.