Albert Reuther

A Comparison of Eleven Static Heuristics for Mapping a Class of Independent Tasks onto Heterogeneous Distributed Computing Systems

Mixed-machine heterogeneous computing (HC) environments utilize a distributed suite of different high-performance machines, interconnected with high-speed links, to perform different computationally intensive applications that have diverse computational requirements. HC environments are well suited to meet the computational demands of large, diverse groups of tasks. The problem of optimally mapping (defined as matching and scheduling) these tasks onto the machines of a distributed HC environment has been shown, in general, to be NP-complete, requiring the development of heuristic techniques. Selecting the best heuristic to use in a given environment, however, remains a difficult problem, because comparisons are often clouded by different underlying assumptions in the original study of each heuristic. Therefore, a collection of 11 heuristics from the literature has been selected, adapted, implemented, and analyzed under one set of common assumptions. It is assumed that the heuristics derive a mapping statically (i.e., off-line). It is also assumed that a metatask (i.e., a set of independent, noncommunicating tasks) is being mapped and that the goal is to minimize the total execution time of the metatask. The 11 heuristics examined are Opportunistic Load Balancing, Minimum Execution Time, Minimum Completion Time, Min?min, Max?min, Duplex, Genetic Algorithm, Simulated Annealing, Genetic Simulated Annealing, Tabu, and A*. This study provides one even basis for comparison and insights into circumstances where one technique will out-perform another. The evaluation procedure is specified, the heuristics are defined, and then comparison results are discussed. It is shown that for the cases studied here, the relatively simple Min?min heuristic performs well in comparison to the other techniques.

A comparison study of static mapping heuristics for a class of meta-tasks on heterogeneous computing systems

Heterogeneous computing (HC) environments are well suited to meet the computational demands of large, diverse groups of tasks (i.e., a meta-task). The problem of mapping (defined as matching and scheduling) these tasks onto the machines of an HC environment has been shown, in general, to be NP-complete, requiring the development of heuristic techniques. Selecting the best heuristic to use in a given environment, however, remains a difficult problem, because comparisons are often clouded by different underlying assumptions in the original studies of each heuristic. Therefore, a collection of eleven heuristics from the literature has been selected, implemented, and analyzed under one set of common assumptions. The eleven heuristics examined are opportunistic load balancing, user-directed assignment, fast greedy, min-min, max-min, greedy, genetic algorithm, simulated annealing, genetic simulated annealing, tabu, and A*. This study provides one even basis for comparison and insights into circumstances where one technique will outperform another. The evaluation procedure is specified, the heuristics are defined, and then selected results are compared.

A taxonomy for describing matching and scheduling heuristics for mixed-machine heterogeneous computing systems

The problem of mapping (defined as matching and scheduling) tasks and communications onto multiple machines and networks in a heterogeneous computing (HC) environment has been shown to be NP-complete, in general, requiring the development of heuristic techniques. Many different types of mapping heuristics have been developed in recent years. However, selecting the best heuristic to use in any given scenario remains a difficult problem. Factors making this selection difficult are discussed. Motivated by these difficulties, a new taxonomy for classifying mapping heuristics for HC environments is proposed (Purdue HC Taxonomy). The taxonomy is defined in three major parts: the models used for applications and communication requests; the models used for target hardware platforms; and the characteristics of mapping heuristics, Each part of the taxonomy is described, with examples given to help clarify the taxonomy. The benefits and uses of this taxonomy are also discussed.

Dynamic distributed dimensional data model (D4M) database and computation system

A crucial element of large web companies is their ability to collect and analyze massive amounts of data. Tuple store databases are a key enabling technology employed by many of these companies (e.g., Google Big Table and Amazon Dynamo). Tuple stores are highly scalable and run on commodity clusters, but lack interfaces to support efficient development of mathematically based analytics. D4M (Dynamic Distributed Dimensional Data Model) has been developed to provide a mathematically rich interface to tuple stores (and structured query language “SQL” databases). D4M allows linear algebra to be readily applied to databases. Using D4M, it is possible to create composable analytics with significantly less effort than using traditional approaches. This work describes the D4M technology and its application and performance.

Characterizing Resource Allocation Heuristics for Heterogeneous Computing Systems

In many distributed computing environments, collections of applications need to be processed using a set of heterogeneous computing (HC) resources to maximize some performance goal. An important research problem in these environments is how to assign resources to applications (matching) and order the execution of the applications (scheduling) so as to maximize some performance criterion without violating any constraints. This process of matching and scheduling is called mapping. To make meaningful comparisons among mapping heuristics, a system designer needs to understand the assumptions made by the heuristics for (1) the model used for the application and communication tasks, (2) the model used for system platforms, and (3) the attributes of the mapping heuristics. This chapter presents a three-part classification scheme ( 3PCS ) for HC systems. The 3PCS is useful for researchers who want to (a) understand a mapper given in the literature, (b) describe their design of a mapper more thoroughly by using a common standard, and (c) select a mapper to match a given real-world environment.

D4M 2.0 schema: A general purpose high performance schema for the Accumulo database

Non-traditional, relaxed consistency, triple store databases are the backbone of many web companies (e.g., Google Big Table, Amazon Dynamo, and Facebook Cassandra). The Apache Accumulo database is a high performance open source relaxed consistency database that is widely used for government applications. Obtaining the full benefits of Accumulo requires using novel schemas. The Dynamic Distributed Dimensional Data Model (D4M)[http://www.mit.edu/~kepner/D4M] provides a uniform mathematical framework based on associative arrays that encompasses both traditional (i.e., SQL) and non-traditional databases. For non-traditional databases D4M naturally leads to a general purpose schema that can be used to fully index and rapidly query every unique string in a dataset. The D4M 2.0 Schema has been applied with little or no customization to cyber, bioinformatics, scientific citation, free text, and social media data. The D4M 2.0 Schema is simple, requires minimal parsing, and achieves the highest published Accumulo ingest rates. The benefits of the D4M 2.0 Schema are independent of the D4M interface. Any interface to Accumulo can achieve these benefits by using the D4M 2.0 Schema.

Achieving 100,000,000 database inserts per second using Accumulo and D4M

The Apache Accumulo database is an open source relaxed consistency database that is widely used for government applications. Accumulo is designed to deliver high performance on unstructured data such as graphs of network data. This paper tests the performance of Accumulo using data from the Graph500 benchmark. The Dynamic Distributed Dimensional Data Model (D4M) software is used to implement the benchmark on a 216-node cluster running the MIT SuperCloud software stack. A peak performance of over 100,000,000 database inserts per second was achieved which is 100× larger than the highest previously published value for any other database. The performance scales linearly with the number of ingest clients, number of database servers, and data size. The performance was achieved by adapting several supercomputing techniques to this application: distributed arrays, domain decomposition, adaptive load balancing, and single-program-multiple-data programming.

Driving big data with big compute

Big Data (as embodied by Hadoop clusters) and Big Compute (as embodied by MPI clusters) provide unique capabilities for storing and processing large volumes of data. Hadoop clusters make distributed computing readily accessible to the Java community and MPI clusters provide high parallel efficiency for compute intensive workloads. Bringing the big data and big compute communities together is an active area of research. The LLGrid team has developed and deployed a number of technologies that aim to provide the best of both worlds. LLGrid MapReduce allows the map/reduce parallel programming model to be used quickly and efficiently in any language on any compute cluster. D4M (Dynamic Distributed Dimensional Data Model) provided a high level distributed arrays interface to the Apache Accumulo database. The accessibility of these technologies is assessed by measuring the effort to use these tools and is typically a few lines of code. The performance is assessed by measuring the insert rate into the Accumulo database. Using these tools a database insert rate of 4M inserts/second has been achieved on an 8 node cluster.

D4M: Bringing associative arrays to database engines

The ability to collect and analyze large amounts of data is a growing problem within the scientific community. The growing gap between data and users calls for innovative tools that address the challenges faced by big data volume, velocity and variety. Numerous tools exist that allow users to store, query and index these massive quantities of data. Each storage or database engine comes with the promise of dealing with complex data. Scientists and engineers who wish to use these systems often quickly find that there is no single technology that offers a panacea to the complexity of information. When using multiple technologies, however, there is significant trouble in designing the movement of information between storage and database engines to support an end-to-end application along with a steep learning curve associated with learning the nuances of each underlying technology. In this article, we present the Dynamic Distributed Dimensional Data Model (D4M) as a potential tool to unify database and storage engine operations. Previous articles on D4M have showcased the ability of D4M to interact with the popular NoSQL Accumulo database. Recently however, D4M now operates on a variety of backend storage or database engines while providing a federated look to the end user through the use of associative arrays. In order to showcase how new databases may be supported by D4M, we describe the process of building the D4M-SciDB connector and present performance of this connection.

LLSuperCloud: Sharing HPC systems for diverse rapid prototyping

The supercomputing and enterprise computing arenas come from very different lineages. However, the advent of commodity computing servers has brought the two arenas closer than they have ever been. Within enterprise computing, commodity computing servers have resulted in the development of a wide range of new cloud capabilities: elastic computing, virtualization, and data hosting. Similarly, the supercomputing community has developed new capabilities in heterogeneous, massively parallel hardware and software. Merging the benefits of enterprise clouds and supercomputing has been a challenging goal. Significant effort has been expended in trying to deploy supercomputing capabilities on cloud computing systems. These efforts have resulted in unreliable, low-performance solutions, which requires enormous expertise to maintain. LLSuperCloud provides a novel solution to the problem of merging enterprise cloud and supercomputing technology. More specifically LLSuperCloud reverses the traditional paradigm of attempting to deploy supercomputing capabilities on a cloud and instead deploys cloud capabilities on a supercomputer. The result is a system that can handle heterogeneous, massively parallel workloads while also providing high performance elastic computing, virtualization, and databases. The benefits of LLSuperCloud are highlighted using a mixed workload of C MPI, parallel MATLAB, Java, databases, and virtualized Web services.