**Êtes-vous un étudiant de l'EPFL à la recherche d'un projet de semestre?**

Travaillez avec nous sur des projets en science des données et en visualisation, et déployez votre projet sous forme d'application sur GraphSearch.

Publication# Reliability Analysis of Data Storage Systems

Résumé

Modern data storage systems are extremely large and consist of several tens or hundreds of nodes. In such systems, node failures are daily events, and safeguarding data from them poses a serious design challenge. The focus of this thesis is on the data reliability analysis of storage systems and, in particular, on the effect of different design choices and parameters on the system reliability. Data redundancy, in the form of replication or advanced erasure codes, is used to protect data from node failures. By storing redundant data across several nodes, the surviving redundant data on surviving nodes can be used to rebuild the data lost by the failed nodes if node failures occur. As these rebuild processes take a finite amount of time to complete, there exists a nonzero probability of additional node failures during rebuild, which eventually may lead to a situation in which some of the data have lost so much redundancy that they become irrecoverably lost from the system. The average time taken by the system to suffer an irrecoverable data loss, also known as the mean time to data loss (MTTDL), is a measure of data reliability that is commonly used to compare different redundancy schemes and to study the effect of various design parameters. The theoretical analysis of MTTDL, however, is a challenging problem for non-exponential real-world failure and rebuild time distributions and for general data placement schemes. To address this issue, a methodology for reliability analysis is developed in this thesis that is based on the probability of direct path to data loss during rebuild. The reliability analysis is detailed in the sense that it accounts for the rebuild times involved, the amounts of partially rebuilt data when additional nodes fail during rebuild, and the fact that modern systems use an intelligent rebuild process that will first rebuild the data having the least amount of redundancy left. Through rigorous arguments and simulations it is established that the methodology developed is well-suited for the reliability analysis of real-world data storage systems. Applying this methodology to data storage systems with different types of redundancy, various data placement schemes, and rebuild constraints, the effect of the design parameters on the system reliability is studied. When sufficient network bandwidth is available for rebuild processes, it is shown that spreading the redundant data corresponding to the data on each node across a higher number of other nodes and using a distributed and intelligent rebuild process will improve the system MTTDL. In particular, declustered placement, which corresponds to spreading the redundant data corresponding to each node equally across all other nodes of the system, is found to potentially have significantly higher MTTDL values than other placement schemes, especially for large storage systems. This implies that more reliable data storage systems can be designed merely by changing the data placement without compromising on the storage efficiency or performance. The effect of a limited network rebuild bandwidth on the system reliability is also analyzed, and it is shown that, for certain redundancy schemes, spreading redundant data across more number of nodes can actually have a detrimental effect on reliability. It is also shown that the MTTDL values are invariant in a large class of node failure time distributions with the same mean. This class includes the exponential distribution as well as the real-world distributions, such as Weibull or gamma. This result implies that the system MTTDL will not be affected if the failure distribution is changed to a corresponding exponential one with the same mean. This observation is also of great importance because it suggests that the MTTDL results obtained in the literature by assuming exponential node failure distributions may still be valid for real-world storage systems despite the fact that real-world failure distributions are non-exponential. In contrast, it is shown that the MTTDL is sensitive to the node rebuild time distribution. A storage system reliability simulator is built to verify the theoretical results mentioned above. The simulator is sufficiently complex to perform all required failure events and rebuild tasks in a storage system, to use real-world failure and rebuild time distributions for scheduling failures and rebuilds, to take into account partial rebuilds when additional node failures occur, and to simulate different data placement schemes and compare their reliability. The simulation results are found to match the theoretical predictions with high confidence for a wide range of system parameters, thereby validating the methodology of reliability analysis developed.

Official source

Cette page est générée automatiquement et peut contenir des informations qui ne sont pas correctes, complètes, à jour ou pertinentes par rapport à votre recherche. Il en va de même pour toutes les autres pages de ce site. Veillez à vérifier les informations auprès des sources officielles de l'EPFL.

Concepts associés

Chargement

Publications associées

Chargement

Concepts associés (35)

Système

Un système est un ensemble d' interagissant entre eux selon certains principes ou règles. Par exemple une molécule, le système solaire, une ruche, une société humaine, un parti, une armée etc.
Un s

Probabilité

vignette|Quatre dés à six faces de quatre couleurs différentes. Les six faces possibles sont visibles.
Le terme probabilité possède plusieurs sens : venu historiquement du latin probabilitas, il désig

Informatique théorique

vignette|Une représentation artistique d'une machine de Turing. Les machines de Turing sont un modèle de calcul.
L'informatique théorique est l'étude des fondements logiques et mathématiques de l'info

Publications associées (124)

Chargement

Chargement

Chargement

Modern industrial, government, and academic organizations are collecting massive amounts of data at an unprecedented scale and pace. The ability to perform timely, predictable and cost-effective analytical processing of such large data sets in order to extract deep insights is now a key ingredient for success. Traditional database systems (DBMS) are, however, not the first choice for servicing these modern applications, despite 40 years of database research. This is due to the fact that modern applications exhibit different behavior from the one assumed by DBMS: a) timely data exploration as a new trend is characterized by ad-hoc queries and a short user interaction period, leaving little time for DBMS to do good performance tuning, b) accurate statistics representing relevant summary information about distributions of ever increasing data are frequently missing, resulting in suboptimal plan decisions and consequently poor and unpredictable query execution performance, and c) cloud service providers - a major winner in the data analytics game due to the low cost of (shared) storage - have shifted the control over data storage from DBMS to the cloud providers, making it harder for DBMS to optimize data access. This thesis demonstrates that database systems can still provide timely, predictable and cost-effective analytical processing, if they use an agile and adaptive approach. In particular, DBMS need to adapt at three levels (to workload, data and hardware characteristics) in order to stabilize and optimize performance and cost when faced with requirements posed by modern data analytics applications. Workload-driven data ingestion is introduced with NoDB as a means to enable efficient data exploration and reduce the data-to-insight time (i.e., the time to load the data and tune the system) by doing these steps lazily and incrementally as a side-effect of posed queries as opposed to mandatory first steps. Data-driven runtime access path decision making introduced with Smooth Scan alleviates suboptimal query execution, postponing the decision on access paths from query optimization, where statistics are heavily exploited, to query execution, where the system can obtain more details about data distributions. Smooth Scan uses access path morphing from one physical alternative to another to fit the observed data distributions, which removes the need for a priori access path decisions and substantially improves the predictability of DBMS. Hardware-driven query execution introduced with Skipper enables the usage of cold storage devices (CSD) as a cost-effective solution for storing the ever increasing customer data. Skipper uses an out-of-order CSD-driven query execution model based on multi-way joins coupled with efficient cache and I/O scheduling policies to hide the non-uniform access latencies of CSD. This thesis advocates runtime adaptivity as a key to dealing with raising uncertainty about workload characteristics that modern data analytics applications exhibit. Overall, the techniques introduced in this thesis through the three levels of adaptivity (workload, data and hardware-driven adaptivity) increase the usability of database systems and the user satisfaction in the case of big data exploration, making low-cost data analytics reality.

The reliability of new overhead electric and telecommunication lines depends principally on the quality of their support structures. These structures are generally made of wood, metal or concrete. The complexity of a natural substance such as wood requires a thorough analysis of the various factors that influence its overall quality. In the case of wood poles, such factors include initial forest growth pattern, the species of wood and its preservative treatment, ageing characteristics, and its various mechanical defects such as knots, cracks etc. The accumulation of knowledge on the effect of the various variables that contribute to the overall quality of a wood support structure permits an optimum use of such a resource. For example, less variability and higher strength of wood support structures permits optimum loading and spacing between structures, thus reducing the number needed in a specific length of an overhead line. If one assumes that in Western Europe 1 wood pole is employed for every 2 inhabitants, and that this proportion increases in less densely populated countries such as the US and Scandinavia, the economics of optimum use of wood as a resource soon become apparent. In less developed countries, the proportions and the economics vary depending on the natural resources such as wood that they employ. The goal of this research is to establish, thanks to non destructive evaluations, a general ageing probabilistic law of the wooden pole based on two distinguished laws: one on the new pole in studying the influence of a grading of the bad elements based on a normal law: "left-truncation of a normal distribution", point 1; and another one based on the in-field wooden pole in exploiting the different parameters such as: the age of the pole, its chemical treatment, its species, its knots etc. in order to define the pole's damage law, point 2. Statistical distribution law of the new wooden pole after grading by non destructive sorting (ultrasounds) of the high mechanical performances supports: This new distribution law is a Gaussian law or evolves to a Log or Weibull's law with 3 parameters according to the inspected species. This grading allows a revalorization of the properties of the new poles and of the design values while guaranteeing an index of reliability required by the design standards, or in improving directly this nominal reliability (economic gain and reliability gain). Statistical distribution law of an aged in-field population (20-50 years old) approached by a bi-modal law which depends on: The distribution law of the new component (see point 1) and its minimal extreme law, which is asymmetrical, for an observation on 50 years. The statistical distribution at the time t of the residual mechanical performances of a group of supports making a local net, evaluated by non destructive methods. The non destructive evaluation is based on the measurements of physical variables (density, biological moisture content) and some descriptive variables from natural origins (diameter, knots, cracks...) and from accidental origins (diameter reduction, lightning cracks...). The statistical distribution at the time t is then obtained on the basis of a model of multivariate non destructive evaluation, generalized to the whole of species and treatments. This model is the other concrete goal to reach in this thesis. As a conclusion, the research demonstrates the influence and the interaction of the new pole grading (distribution at t0) on the modelisation of the distribution at ti (multivariate non destructive model). The data used for the mentioned modelisations come from a significant international database with a large amount of inspected wood poles and with studied cases. This database is the synthesis of about 15 years of research and development leaded by IBOIS-EPFL and its international partners. The probabilistic approaches are then validated by a huge database allowing thus to be directly exploitable. On this basis, all the standards dealing with the new poles and dealing with the controls and maintenances of a wooden pole networks, could be re-examined for a double gain: Concerning the economy: by increasing the capacity of the new poles profiting of an objective quality assurance, and by increasing the life time of the in-field pole, in knowing how to purge only the ones which are under the critical threshold of damage Concerning the reliability: by increasing the reliability of the network from the stage "new pole", by eliminating the weakest components, and by maintaining this reliability during all the life time of the network thanks to a cyclic preventive maintenance (every 5 to 8 years) and the replacement of only the weakened poles.

Christina Fragouli, Robert Haas, Xiao-Yu Hu, Vinodh Venkatesan

Replication is a widely used method to protect large- scale data storage systems from data loss when storage nodes fail. It is well known that the placement of replicas of the different data blocks across the nodes affects the time to rebuild. Several systems described in the literature are designed based on the premise that minimizing the rebuild times maximizes the system reliability. Our results however indicate that the reliability is essentially unaffected by the replica placement scheme. We show that, for a replication factor of two, all possible placement schemes have mean times to data loss (MTTDLs) within a factor of two for practical values of the failure rate, storage capacity, and rebuild bandwidth of a storage node. The theoretical results are confirmed by means of event-driven simulation. For higher replication factors, an analytical derivation of MTTDL becomes intractable for a general placement scheme. We therefore use one of the alternate measures of reliability that have been proposed in the literature, namely, the probability of data loss during rebuild in the critical mode of the system. Whereas for a replication factor of two this measure can be directly translated into MTTDL, it is only speculative of the MTTDL behavior for higher replication factors. This measure of reliability is shown to lie within a factor of two for all possible placement schemes and any replication factor. We also show that for any replication factor, the clustered placement scheme has the lowest probability of data loss during rebuild in critical mode among all possible placement schemes, whereas the declustered placement scheme has the highest probability. Simulation results reveal however that these properties do not hold for the corresponding MTTDLs for a replication factor greater than two. This indicates that some alternate measures of reliability may not be appropriate for comparing the MTTDL of different placement schemes.

2010