Section: New Results

Convergence of HPC and Big Data

Týr: Blob-based storage convergence of HPC and Big Data

Participants : Pierre Matri, Alexandru Costan, Gabriel Antoniu.

The increasingly growing data sets processed on HPC platforms raise major challenges for the underlying storage layer. A promising alternative to POSIX-I/O-compliant file systems are simpler blobs (binary large objects), or object storage systems. They offer lower overhead, better performance and horizontal scalability at the cost of largely unused features such as file hierarchies or permissions. Similarly, blobs are increasingly considered for replacing distributed file systems for big data analytics or as a base for storage abstractions like key-value stores or time-series databases.

This growing interest from both HPC and Big Data communities towards blob storage naturally fits with the current trend towards HPC and Big Data convergence. In this context, we seek to demonstrate that blob storage indeed constitutes a strong alternative to current storage infrastructures. Additionally, the data model of blob storage is close enough to that of distributed file systems so that this change is largely transparent for the applications running atop them.

In [22] we provide a preliminary evaluation of blob storage in HPC and Big Data contexts. We leverage a series of real-world HPC applications as well as an industry-standard HPC benchmark. We analyze for each of these applications the storage requests sent to the underlying storage system. We discover that over 98% of these storage calls can be directly mapped to the data model offered by blobs. Interestingly, we also note that the remaining calls are using file systems features for convinience rather than by necessity. These calls may consequently be performed as offline pre- or post-processing, or avoided altogether without altering the application.

Modeling elastic storage

Participants : Nathanaël Cheriere, Gabriel Antoniu.

For efficient Big Data processing, efficient resource utilization becomes a major concern as large-scale computing infrastructures such as supercomputers or clouds keep growing in size. Naturally, energy and cost savings can be obtained by reducing idle resources. Malleability, which is the possibility for resource managers to dynamically increase or reduce the resources of jobs, appears as a promising means to progress towards this goal.

However, state-of-the-art parallel and distributed file systems have not been designed with malleability in mind. This is mainly due to the supposedly high cost of storage decommission, which is considered to involve expensive data transfers. Nevertheless, as network and storage technologies evolve, old assumptions on potential bottlenecks can be revisited.

In [18], we evaluate the viability of malleability as a design principle for a distributed file system. We specifically model the duration of the decommission operation, for which we obtain a theoretical lower bound. Then we consider HDFS as a use case and we show that our model can explain the measured decommission times.

The existing decommission mechanism of HDFS is good when the network is the bottleneck, but could be accelerated by up to a factor 3 when the storage is the limiting factor. With the highlights provided by our model, we suggest improvements to speed up decommission in HDFS and we discuss open perspectives for the design of efficient malleable distributed file systems.

Eley: Leveraging burst-buffers for efficient Big Data processing on HPC systems

Participants : Orçun Yildiz, Chi Zhou, Shadi Ibrahim.

Burst Buffer is an effective solution for reducing the data transfer time and the I/O interference in HPC systems. Extending Burst Buffers (BBs) to handle Big Data applications is challenging because BBs must account for the large data inputs of Big Data applications and the performance guarantees of HPC applications – which are considered as first-class citizens in HPC systems. Existing BBs focus on only intermediate data of Big Data applications and incur a high performance degradation of both Big Data and HPC applications.In [26], we present Eley, a burst buffer solution that helps to accelerate the performance of Big Data applications while guaranteeing the performance of HPC applications. In order to improve the performance of Big Data applications, Eley employs a prefetching technique that fetches the input data of these applications to be stored close to computing nodes thus reducing the latency of reading data inputs. Moreover, Eley is equipped with a full delay operator to guarantee the performance of HPC applications – as they are running independently on a HPC system. The experimental results show the effectiveness of Eley in obtaining shorter execution time of Big Data applications (shorter map phase) while guaranteeing the performance of HPC applications.