Just checked out the papers for this year’s FAST ’10 in San Jose this week. They are impressive.
Here are a few that caught my eye with quotes, mostly from the abstracts, but sometimes from the results.
SRCMap: Energy Proportional Storage using Dynamic Consolidation
During a given consolidation interval, SRCMap activates a minimal set of physical volumes to serve the workload and spins down the remaining volumes, redirecting their workload to replicas on active volumes. We present both theoretical and experimental evidence to establish the effectiveness of SRCMap in minimizing the power consumption of enterprise storage systems.
Evaluating Performance and Energy in File System Server Workloads
We concluded that default file system types and options are often suboptimal: simple changes within a file system, like mount options, can improve power/performance from 5% to 149%; and chang-ing format options can boost the efficiency from 6% to 136%. Switching to a different file system can result in improvements ranging from 2 to 9 times.
Provenance for the Cloud
While it is feasible to provide provenance as a layer on top of today’s cloud offerings, we conclude by presenting the case for incorporating provenance as a core cloud feature, discussing the issues in doing so.
Discovery of Application Workloads from Network File Traces
Our method is successful at discovering the application-level behavioral characteristics from NFS traces. We have also shown that given a long sequence of NFS trace headers, it is able to annotate regions of the sequence as belonging to the applications that it has been trained with. It can identify and annotate both sequential and concurrent execution of different workloads. Finally, we demonstrate that small snippets of traces are sufficient for identifying many workloads.
BASIL: Automated IO Load Balancing Across Storage Devices
In this paper, we introduce BASIL, a novel software system that automatically manages virtual disk placement and performs load balancing across devices without assuming any support from the storage arrays. BASIL uses IO latency as a primary metric for modeling.
Panache: A Parallel File System Cache for Global File Access
Panache is a scalable, high-performance, clustered file system cache for parallel data-intensive applications that require wide area file access. Panache is the first file system cache to exploit parallelism in every aspect of its design—parallel applications can access and update the cache from multiple nodes while data and metadata is pulled into and pushed out of the cache in parallel. Data is cached and updated using pNFS, which performs parallel I/O between clients and servers, eliminating the single-server bottleneck of vanilla client-server file access protocols. Furthermore, Panache shields applications from fluctuating WAN latencies and outages and is easy to deploy as it relies on open standards for high-performance file serving and does not require any proprietary hardware or software to be installed at the remote cluster.
Accelerating Parallel Analysis of Scientific Simulation Data via Zazen
We have implemented our methodology in a parallel disk cache system called Zazen. By avoiding the overhead associated with querying metadata servers and by reading data in parallel from local disks, Zazen is able to deliver a sustained read bandwidth of over 20 gigabytes per second on a commodity Linux cluster with 100 nodes, approaching the optimal aggregated I/O bandwidth attainable on these nodes. Compared with conventional NFS, PVFS2, and Hadoop/HDFS, respectively, Zazen is 75, 18, and 6 times faster for accessing large (1-GB) files, and 25, 13, and 85 times faster for accessing small (2-MB) files.
Write Endurance in Flash Drives: Measurements and Analysis
Our chip-level measurements show endurance far in excess of nominal values quoted by manufacturers, by a factor of as much as 100. We reverse engineer specifics of the Flash Translation Layers (FTLs) used by several devices, and find a close correlation between measured whole-device endurance and predictions from reverse-engineered FTL parameters and measured chip endurance values.
Extending SSD Lifetimes with Disk-Based Write Caches
In this paper, we propose Griffin, a hybrid storage design that, somewhat contrary to intuition, uses a hard disk drive to cache writes to an SSD. Writes to Griffin are logged sequentially to the HDD write cache and later migrated to the SSD. Reads are usually served from the SSD and occasionally from the slower HDD. Griffin’s goal is to minimize the writes sent to the SSD without significantly impacting its read performance; by doing so, it conserves erase cycles and extends SSD lifetime.
DFS: A File System for Virtualized Flash Storage
Instead of using traditional layers of abstraction, our layers of abstraction are designed for directly accessing flash memory devices. DFS has two main novel features. First, it lays out its files directly in a very large virtual storage address space provided by FusionIO’s virtual flash storage layer. Second, it leverages the virtual flash storage layer to perform block allocations and atomic updates. As a result, DFS performs better and it is much simpler than a traditional Unix file system with similar functionalities.
Understanding latent sector errors and how to protect against them
The statistical analysis revealed some interesting properties. We observe that many of the statistical aspects of LSEs are well modeled by power-laws, including the length of error bursts (i.e. a series of contiguous sectors affected by LSEs), the number of good sectors that separate error bursts, and the number of LSEs observed per time.
We find that these properties are poorly modeled by the most commonly used distributions, geometric and Poisson. Instead we observe that a Pareto distribution fits the data very well and report the parameters that provide the best fit. We hope this data will be useful for other researchers who do not have access to field data. We find no significant difference in the statistical properties of LSEs in nearline drives versus enterprise class drives.
Some of our statistical observations might also hold some clues as to what mechanisms cause LSEs. For example, we observe that nearly all drives with LSEs, experience all LSEs in their lifetime within the same 2-week period, indicating that for most drives most errors have been caused by the same event (e.g. one scratch), rather than a slow and continuous wear-out of the media.
A Clean-Slate Look at Disk Scrubbing
Our work is a first step in the exploration of more intelligent scrubbing strategies for hard drives. It shows that single drive reliability can be greatly improved by expanding the design space for scrubbing strategies be- yond naive sequential and constant-rate approaches.
The StorageMojo take
Wow. Power, flash, errors and error handling, advanced file systems, virtualization, there’s something for everyone. It reads like tomorrow’s news.
The competition for best paper is intense this year – and that’s a good thing. I’ll be attending the conference and look forward to writing more about some of these papers.
Courteous comments welcome, of course. IIRC, the papers will be available on the FAST web site on the 24th.