David D. Chambliss

Performance virtualization for large-scale storage systems

Current data centers require storage capacities of hundreds of terabytes to petabytes. Time-critical applications such as online transactions processing depend on getting adequate performance from the storage subsystem: otherwise, they fail. It is difficult to provide predictable quality of service at this level of complexity, because I/O workloads are extremely variable and device behavior is poorly understood. Ensuring that unrelated but competing workloads do not affect each others performance is still more difficult, and equally necessary. We present SLEDS (Service Level Enforcement Discipline for Storage), a distributed controller that provides statistical performance guarantees on a storage system built from commodity components. SLEDS can adaptively handle unpredictable workload variations so that each client continues to get the performance it needs even in the presence of misbehaving, competing peers. After evaluating the SLEDS on a heterogeneous mid-range storage system, we found that it is vastly superior to the raw system in its ability to provide performance guarantees, while only introducing a negligible overhead.

Ordered nucleation of Ni and Au islands on Au(111) studied by scanning tunneling microscopy

The scanning tunneling microscope reveals that Ni deposited on Au(111) at room temperature forms regular arrays of two‐dimensional islands. The islands grow with spacing 73 A in rows 140 A apart at sites determined by the Au(111) ‘‘herringbone’’ reconstruction. This nucleation at evenly spaced sites yields islands with a narrow size distribution. The apparent Ni island height (1.9 A) is bias‐independent and agrees with a hard‐sphere model of pseudomorphic Ni/Au(111). The behavior of Ni is contrasted with Au deposited on Au(111), for which far fewer islands are formed.

Nucleation and growth of ultrathin Fe and Au films on Cu(100) studied by scanning tunneling microscopy

Cu(100) surfaces modified by depositing Au and Fe in the monolayer (ML) thickness regime are studied with the scanning tunneling microscope (STM). STM results confirm that Au deposition of 0.5 ML at 300 K creates a CuAu c(2×2) alloy monolayer, with nearly regular disruptions of structure that relieve misfit strain. The alloy forms by Au replacement of Cu atoms in the surface. Submonolayer Fe deposition at 300 K yields monolayer islands and additional features interpreted as Fe patches in the surface Cu layer. This suggests that Fe atoms replace Cu atoms in the surface. Deposition of ∼4 ML Fe yields a more homogeneous surface with height variations that suggest structural disorder in the Fe layer. Fe deposition at ∼110 K and warming to 300 K yields many monolayer and bilayer islands, with no evidence of Fe incorporation into the surface. Unusual tip behavior is also discussed which appears to yield STM sensitivity to the chemical difference between Fe and Cu areas.

Growth and morphology of partial and multilayer Fe thin films on Cu(100) and the effect of adsorbed gases studied by scanning tunneling microscopy

Fe epitaxy on Cu(100) is investigated for Fe coverages θ≤3.0 ML. The layer filling statistics are quantitatively related to an evolving growth process, which includes intermixing at the substrate overlayer interface. The resulting inhomogeneous substrate surface and first layer affect the processes by which arriving Fe atoms add to the growth front. Our results explain the previously reported covering of initial Fe not as bilayer growth, but instead as the result of island growth on top of Fe incorporated in the top substrate layer. First layer composition and structure influence the nucleation and growth of the second layer. Island coalescence and formation of a first layer percolation network change the connected first layer area thereby changing the nucleation and growth behavior of the second layer. After both first and second layer growth are completed, images show additional growth is much more layer‐by‐layer in nature. Oxygen exposure after Fe deposition changes the layer filling by both promotion ...

Surface alloy formation studied by scanning tunneling microscopy : Cu(100) + Au-c(2×2)

Abstract The ordered surface alloy formed by Au deposited on Cu(100) at room temperature is studied using scanning tunneling microscopy (STM). The formation of a CuAu alloy monolayer is demonstrated by STM images showing c(2 × 2) ordering with two unequal peaks per unit cell. Longer-range images show linear features in which misfit strain is apparently relieved by transverse atomic displacements. The surface topography of islands shows that they also consist of a CuAu alloy layer, which demonstrates that alloy formation arises by Au replacement of Cu atoms in the surface.

Mixing Deduplication and Compression on Active Data Sets

Many new storage systems provide some form of data reduction. We examine data reduction methods that might be suitable for \emph{primary} storage systems serving active data (as contrasted with backup and archive systems), by analysis of file sets found in different active data environments. We address questions of: how effective are compression and variations of deduplication, both separately and in combination, when deduplication and compression are combined, which should be applied first, what will the tradeoff be between the different methods in their use of MIPS relative to the data reduction achieved, and what degree of data reduction should be expected for different data types.

Surface structure and metal epitaxy: STM studies of ultrathin metal films on Au(111) and Cu(100)

Abstract The scanning tunneling microscope (STM) is an important tool for studying the growth of ultrathin metal structures. The behavior of atoms arriving at the surface determines nanometer-scale structure that is readily measured with the STM. These structural features are important in determining properties. The variety of structural possibilities is illustrated with the difference between substrate-controlled island nucleation of Ni on Au(111) and diffussion-controlled aggregation of Ag on Au(111). The STM also provides a fairly complete picture of the intermixing that occurs in the early stages of room-temperature growth of Fe on Cu(100).

A structural model and mechanism for Fe epitaxy on Cu(100)

A structural and mechanistic model for initial room temperature Fe epitaxy on Cu(100) is presented, based on scanning tunneling microscopy data. Changes in Fe atom attachment kinetics with coverage θ yield several growth regimes: Fe incorporation into the surface by atomic exchange with Cu (θ < 0.2), growth of first-layer Fe islands (0.2 < θ < 0.7), and simultaneous layer-1 and layer-2 growth (0.7 < θ < 2). These results reconcile qualitative disparities in previous interpretations of experimental results.

Insights for data reduction in primary storage: a practical analysis

There has been increasing interest in deploying data reduction techniques in primary storage systems. This paper analyzes large datasets in four typical enterprise data environments to find patterns that can suggest good design choices for such systems. The overall data reduction opportunity is evaluated for deduplication and compression, separately and combined, then in-depth analysis is presented focusing on frequency, clustering and other patterns in the collected data. The results suggest ways to enhance performance and reduce resource requirements and system cost while maintaining data reduction effectiveness. These techniques include deciding which files to compress based on file type and size, using duplication affinity to guide deployment decisions, and optimizing the detection and mapping of duplicate content adaptively when large segments account for most of the opportunity.

Using STM to understand diffraction oscillations for Fe growth on Cu(100)

Abstract The relationship between diffraction intensity variations and topography measured with the scanning tunneling microscope (STM) is examined quantitatively, for room-temperature growth of Fe on Cu(100). Predictions from STM data based on kinematical formulas yield good agreement with the experimental medium-energy electron diffraction (MEED) results of Thomassen et al. [Surf. Sci. 264 (1992) 406]. The agreement demonstrates the similarity of samples prepared in the different laboratories, the applicability of kinematical analysis to MEED oscillations, and the importance of identifying and understanding the characteristic lengths of the sample and the measurement.