Data storage technology today faces many challenges, including performance
inefficiencies, inadequate dependability and integrity guarantees, limited scalability, loss of confidentiality,
poor resource sharing, and increased ownership and management costs. IOLanes examines how multicore
CPUs can be used to design next generation persistent I/O architectures. The project addresses overheads
and scalability issues over multicore CPUs that will allow I/O performance to scale as well as take advantage
of additional compute cycles in future multicores. By improving scaling of the I/O stack on modern
architectures, the project will result in significant increase in overall I/O efficiency: IOLanes designs I/O
subsystems that will be capable of sustaining one order of magnitude more load per storage node than
todays’ systems, resulting in a 10x reduction in the number of storage nodes in data centres or a 10x
increase in the size of the workload for the same number of storage nodes. Considerable consolidation of
resources in data centres and other IT facilities will contribute to more efficient processing of data-centric
applications and a greener world. Users of data-oriented services can expect a cost reduction for their
services as well as new features that will increase productivity.
IOLanes targets three major challenges: (i) dealing with performance and scalability issues of the I/O stack
on multicore architectures, (ii) addressing I/O performance and dynamic resource management issues in
virtualised, single-host environments, and (iii) examining on-loading and off-loading trade-offs for advanced
functions that are becoming essential in modern storage systems, e.g. compression, protection, encryption,
error correction.
IOLanes aims at analysing and addressing these challenges throughout the I/O path. It breaks down the I/
O stack in four important layers: (a) application and middleware, (b) virtual machine, (c) host operating
system, and (d) embedded storage controller. The proposed work analyses and addresses the inefficiencies
associated with these layers on multicore CPUs, by designing an I/O stack that minimizes unnecessary
overheads and scales with the number of cores. IOLanes designs and builds a new I/O stack in the Linux
kernel that achieves partitioning of shared resources and scales with the number of cores. This stack
supports virtualized I/O and offers facilities for transparently using the available cycles in multicore
processors for new I/O features. In addition, IOLanes builds a single-point monitoring infrastructure for
complex, persistent I/O stacks. The project will demonstrate the benefits of the new design using real-life
datacentre-type applications.
The IOLANES project is supported by partners Foundation for Research and Technology – Hellas (FORTH)
- Coordinator, Barcelona Supercomputing Centre (BSC), University of Madrid (UPM), INTEL Performance
Labs Ireland, IBM Research Labs in Haifa, and Neurocom S.A. IOLANES is funded by the E.C. under the 7th
Framework program and is part of the portfolio of the Embedded Systems Unit - G3 Directorate General
Information Society.
Contact:
Prof. Angelos Bilas
Foundation for Research and Technology – Hellas (FORTH)
Institute of Computer Science (ICS), www.ics.forth.gr
N. Plastira 100, GR-700 13, Heraklion, Crete, Greece
Tel: +30 2810391669, Fax: +30 2810391661, Email: bilas@ics.forth.gr