In-memory Data Structure Research Articles

Scalable Time-Range k -Core Query on Temporal Graphs

Querying cohesive subgraphs on temporal graphs with various time constraints has attracted intensive research interests recently. In this paper, we study a novel Temporal k -Core Query (TCQ) problem: given a time interval, find all distinct k -cores that exist within any subintervals from a temporal graph, which generalizes the previous historical k -core query. This problem is challenging because the number of subintervals increases quadratically to the span of time interval. For that, we propose a novel Temporal Core Decomposition (TCD) algorithm that decrementally induces temporal k -cores from the previously induced ones and thus reduces "intra-core" redundant computation significantly. Then, we introduce an intuitive concept named Tightest Time Interval (TTI) for temporal k -core, and design an optimization technique with theoretical guarantee that leverages TTI as a key to predict which subintervals will induce duplicated k -cores and prunes the subintervals completely in advance, thereby eliminating "inter-core" redundant computation. The complexity of optimized TCD (OTCD) algorithm no longer depends on the span of query time interval but only the scale of final results, which means OTCD algorithm is scalable. Moreover, we propose a compact in-memory data structure named Temporal Edge List (TEL) to implement OTCD algorithm efficiently in physical level with bounded memory requirement. TEL organizes temporal edges in a "timeline" and can be updated instantly when new edges arrive in dynamical temporal graphs. We compare OTCD algorithm with the incremental historical k -core query on several real-world temporal graphs, and observe that OTCD algorithm outperforms it by three orders of magnitude, even though OTCD algorithm needs none precomputed index.

VIP hashing

All data is not equally popular. Often, some portion of data is more frequently accessed than the rest, which causes a skew in popularity of the data items. Adapting to this skew can improve performance, and this topic has been studied extensively in the past for disk-based settings. In this work, we consider an in-memory data structure, namely hash table , and show how one can leverage the skew in popularity for higher performance. Hashing is a low-latency operation, sensitive to the effects of caching and code complexity, among other factors. These factors make learning in-the-loop challenging as the overhead of performing additional operations can have significant impact on performance. In this paper, we propose VIP hashing, a hash table method that uses lightweight mechanisms for learning the skew in popularity and adapting the hash table layout on the fly. These mechanisms are non-blocking, i.e, the hash table is operational at all times. The overhead is controlled by sensing changes in the popularity distribution to dynamically switch-on/off the mechanisms as needed. We ran extensive tests against a host of workloads generated by Wiscer , a homegrown benchmarking tool, and we find that VIP hashing improves performance in the presence of skew (22% increase in fetch operation throughput for a hash table with 1M keys under low skew) while adapting to insert and delete operations, and changing popularity distribution of keys on the fly. Our experiments on DuckDB show that VIP hashing reduces the end-to-end execution time of TPC-H query 9 by 20% under low skew.

Mille Cheval: a GPU-based in-memory high-performance computing framework for accelerated processing of big-data streams

Streams are temporally ordered, rapid changing, ample in volume, and infinite in nature. It is nearly impossible to store the entire data stream due to its large volume and high velocity. In this work, the principle of parallelism is employed to accelerate stream data computing. GPU-based high-performance computing (HPC) framework is proposed for accelerated processing of big-data streams using the in-memory data structure. We have implemented three parallel algorithms to prove the viability of the framework. The contributions of Mille Cheval are: (1) the viability of streaming on accelerators to increase throughput, (2) carefully chosen hash algorithms to achieve low collision rate and high randomness, and (3) memory sketches for approximation. The objective is to leverage the power of a single node using in-memory computing and hybrid computing. HPC does not always require high-end hardware but well-designed algorithms. Achievements of Mille Cheval are: (1) relative error is 1.32 when error rate and overestimate rate are chosen as 0.001 and (2) the host memory space requirement is just 63 MB for 1 terabyte of data. The proposed algorithms are pragmatic. It is evident from experimental results that the framework demonstrates 10X speed-up as compared with CPU implementations and 3X speed-up as compared with GPU implementations.

A GPU-Accelerated In-Memory Metadata Management Scheme for Large-Scale Parallel File Systems

Driven by the increasing requirements of high-performance computing applications, supercomputers are prone to containing more and more computing nodes. Applications running on such a large-scale computing system are likely to spawn millions of parallel processes, which usually generate a burst of I/O requests, introducing a great challenge into the metadata management of underlying parallel file systems. The traditional method used to overcome such a challenge is adopting multiple metadata servers in the scale-out manner, which will inevitably confront with serious network and consistence problems. This work instead pursues to enhance the metadata performance in the scale-up manner. Specifically, we propose to improve the performance of each individual metadata server by employing GPU to handle metadata requests in parallel. Our proposal designs a novel metadata server architecture, which employs CPU to interact with file system clients, while offloading the computing tasks about metadata into GPU. To take full advantages of the parallelism existing in GPU, we redesign the in-memory data structure for the name space of file systems. The new data structure can perfectly fit to the memory architecture of GPU, and thus helps to exploit the large number of parallel threads within GPU to serve the bursty metadata requests concurrently. We implement a prototype based on BeeGFS and conduct extensive experiments to evaluate our proposal, and the experimental results demonstrate that our GPU-based solution outperforms the CPU-based scheme by more than 50% under typical metadata operations. The superiority is strengthened further on high concurrent scenarios, e.g., the high-performance computing systems supporting millions of parallel threads.

NVGraph: Enforcing Crash Consistency of Evolving Network Analytics in NVMM Systems

Many complex networks can be modeled as evolving graphs. Existing in-memory graph data structures are designed for DRAM. Thus, they cannot effectively exploit the current and ongoing adoption of emerging non-volatile main memory (NVMM) for two reasons. (1) Ephemeral graph data structures are not crash-consistent for NVMM. (2) NVMM writes and reads and may incur higher latency than DRAM. In this article, we propose a novel persistent evolving graph data structure, named NVG RAPH , for both computing and in-memory storage of evolving graphs in NVMM. We devise NVG RAPH as a multi-version data structure, wherein a minimum of one version of its data is stored in NVMM to provide the desired durability at runtime for failure recovery, and another version is stored in both DRAM and NVMM to reduce the NVMM-induced memory latency. NVG RAPH is also a partitioned data structure. We dynamically transform the layout of NVG RAPH (e.g., changing the size of its partition in DRAM) exploiting network properties and data access patterns of workloads. For the evaluation of NVG RAPH , we implement four representative real-world graph applications: pagerank, BFS, influence maximization, and rumor source detection. The experimental results show that the performance of NVG RAPH is comparable to other in-memory data structure (e.g., CSR and LLAMA) while using 70 percent less DRAM. It scales well up to 10 billion edges and 201 snapshots and supports crash consistency. It offers up to the 21X speedup of execution time compared to the scale-up graph computation approaches (e.g., GraphChi and X-stream).

TeksDB

Key-value stores (KVS) are now an integral part of modern dataintensive systems. thanks to its simplicity, scalability, and efficiency over traditional database systems. Databases such as MySQL employ KVS (in this case, RocksDB as their backend storage instead of conventional database engines because KVS is less affected by data fragmentation. OLAP and big data analytics applications often use KVS as the backend store to make use of the efficient range_query operations of LSM(Log Structured Merge Tree)-based KVS. Furthermore, the increasing use of KVS as backend stores in distributed systems diversifies the workload for each local KVS, making it necessary to design a highly-concurrent data store that simultaneously processes mixed operations. The underlying technology trends also put immense pressure on the design of KVS. Nowadays, many core servers with hundred GBs of memory are common in production environments, and this enhanced density of hardware technologies (both in terms of the number of cores and memory capacity) leads to more data items residing in memory. This phenomenon places greater emphasis on the efficiency of the in-memory data structure in KVS, as more data accesses are serviced from its memory component, which is often called the memtable. The current memtable structure of KVS, however, generally assumes a limited memory capacity that only serves as a temporary write buffer, and the use of single-threaded writes or unscalable data structures for the memtable fails to fulfill this external demand.

Pulsating STM – The in-memory Optimistic Concurrency Control Technique for Multi Core Systems

In the world of ever increasing parallelism, the problem of deadlock-free concurrency control is inevitable. As the number of processing cores is increasing, the number of processing threads is also increasing, and with this increase in the number of processing threads, there is a good chance of problems arising due to lack of proper concurrency control. The application areas under the domain of advanced graphics, cryptography, deep learning, embedded system programming, artificial intelligence and networking are prone to the problems of heavy uncontrolled concurrency of threads. This paper presents a novel Software Transactional Memory (STM) based optimistic concurrency control technique that is deadlock free for threads accessing the in-memory data structure for the purpose of reading as well as writing. The technique is lock free and is based upon timestamping. Threads involved in the proposed approach possess the transactional properties of atomicity, concurrency and isolation. Durability is not expected as the threads are working on an in-memory data source. The approach involves lazy conflict detection that ensures minimum aborts and restarts as well as maximum concurrency among transactions. Being lock free, the algorithm is better than the existing lock-based techniques. The technique is tested on Sniper-6.1 multi core simulator simulating 64 CPU cores and running 16, 32, 40 and 50 threads in our case. The results show significant improvement in throughput with the increasing number of threads over the existing lock-based techniques as well as other STM techniques based on optimistic concurrency control.

TeksDB

In this paper, we examine the design tradeoffs of existing in-memory data structures of a state-of-the-art key-value store. We observe that no data structures provide both fast point-accesses and consistent ranged- retrievals, and naive amalgamations of existing structures fail to get the best of both worlds. Furthermore, our experiments reveal a performance anomaly when increasing the memory size: as more key-value pairs are maintained in memory, the shortcomings of the data structures exacerbate. To address the above problems, we present TeksDB, a fast and consistent key-value store with a novel in-memory data structure, which effciently handles both point- and ranged- accesses at a modest increase in memory footprint. Our evaluation demonstrates that TeksDB outperforms RocksDB by 3.6×, 9×, and 4.5× for get, scan, and range_query, respectively. The effectiveness of TeksDB extends to real-world workloads, achieving up to 3.3× speedup for YCSB.

Improving the Scalability and Replicability of Embedded Systems Remote Laboratories Through a Cost-Effective Architecture

Online remote laboratories are a particularly promising tool for effective STEM education. They offer online universal access to different hardware devices in which students can experiment and can test and improve their knowledge. However, most of them have two significant limitations. First, given that most of them are developed as, or evolve from single-user proofs of concept, they have no scalability provisions other than full laboratory replication. And second, when this is done, cost efficiency is often neglected. This paper presents the requirements for the creation of a novel remote laboratory architecture focused on, but not limited to, embedded systems experimentation. An architecture, based on Redis (an open source, in-memory data structure store, which is often used as database, cache or message broker), a modular design, and hardware-sharing techniques, is proposed in order to achieve the combined requirements of high scalability and cost efficiency. This mixed hardware-software architecture serves as a basis for the development of remote laboratories, especially those focused on microcontroller-based systems experimentation and embedded devices experimentation. From a user perspective the architecture is web-based, and has provisions to be easily adaptable to different Learning Management Systems and different hardware embedded devices. A new microcontroller-oriented remote laboratory based on the architecture has been developed, with the aim of providing valid evaluation data, and has been used in a real environment. The architecture and the resulting remote laboratory have been compared with other state of the art remote laboratories and their architectures. Results suggest that the proposed architecture does indeed meet the main requirements, which are scalability through replicability and cost efficiency. Furthermore, similarly to previous architectures, it promotes usability, universal access, modularity and reliability.

Enhanced Indexing and Querying of Trajectories in Road Networks via String Algorithms

In this article, we propose a novel indexing and querying method for trajectories constrained in a road network. We aim to provide efficient algorithms for various types of spatiotemporal queries that involve routing in road networks, such as (1) finding moving objects that have traveled along a given path during a given time interval, (2) extracting all paths traveled after a given spatiotemporal context, and (3) enumerating all paths between two locations traveled during a certain time interval. Unlike the existing methods in spatial database research, we employ indexing techniques and algorithms from string processing. This idea is based on the fact that we can represent spatial paths as strings, because trajectories in a network are represented as sequences of road segment IDs. The proposed SNT-index (<u>s</u>uffix-array-based <u>n</u>etwork-constrained <u>t</u>rajectory index) introduces two novel concepts to trajectory indexing. The first is FM-index, which is a compact in-memory data structure for pattern matching. The second is an inverse suffix array, which allows the FM-index to be integrated with the temporal information stored in a forest of B + -trees. Thanks to these concepts, we can reduce the number of B + -tree accesses required by the query processing algorithms to a constant number, something that cannot be achieved with existing methods. Although an FM-index is essentially a static index, we also propose a practical method of appending new data to the index. Finally, experiments show that our method can process the target queries for more than 1 million trajectories in a few tens of milliseconds, which is significantly faster than what the baseline algorithms can achieve without string algorithms.

A Real-Time Algorithm for Tracking Astray Pilgrim based on in-Memory Data Structures

Large crowd management presents a significant challenge to organizers and for the success of the event and to achieve the set objectives. One of the biggest events and with largest crowd in the world is the Muslim pilgrimage to Mecca that happens every year and lasts for five years. The event hosts over two million people from over 80 countries across the world with men, women, and children of various age groups and many languages. One of the challenges that faces the authorities in Saudi Arabia is that many of the pilgrims become astray during the event due to the relative complexity of the rituals mainly mountainous landscape and the language barrier. This result in them being unable to perform the required rituals on the prescribed time(s) with the possibility to invalidate the whole pilgrimage and jeopardize their once-in-a-life journey. Last year over 20,000 pilgrims went astray during the pilgrimage season. In this paper we present a tracking algorithm to help track, alarm, and report astray pilgrims. The algorithm is implemented on a server that contains pilgrims’ data such as geolocations, time stamp and personal information such as name, age, gender, and nationality. Each pilgrim is equipped with a wearable device to report the geolocations and the timestamp to the centralized server. Pilgrims are organized in groups of 20 persons at maximum. By identifying the distance of the pilgrim to its group’s centroid and whether or not the pilgrim’s geolocation is where it is supposed to be according to the pilgrimage schedule, the algorithm determines if the pilgrim is astray or on a verge of becoming astray. Algorithm complexity analysis is performed. For better performance and shorter real-time time to determine the pilgrim’s status, the algorithm employs an in-memory data structure. The analysis showed that the time complexity is O(n). The algorithm has also been tested using simulation runs based on synthesized data that is randomly generated within a specified geographical zone and according to the pilgrimage plan. The simulation results showed good agreement with the analytical performance analysis.

MatrixMap: Programming abstraction and implementation of matrix computation for big data analytics

The computation core of many big data applications can be expressed as general matrix computations, including linear algebra operations and irregular matrix operations. However, existing parallel programming systems such as Spark do not have programming abstraction and efficient implementation for general matrix computations. In this paper, we present MatrixMap, a unified and efficient data-parallel programming framework for general matrix computations. MatrixMap provides powerful yet simple abstraction, consisting of a distributed in-memory data structure called bulk key matrix and a programming interface defined by matrix patterns. Users can easily load data into bulk key matrices and program algorithms into parallel matrix patterns. MatrixMap outperforms current state-of-the-art systems by employing three key techniques:matrix patterns with lambda functions for irregular and linear algebra matrix operations, asynchronous computation pipeline with context-aware data shuffling strategies for specific matrix patterns and in-memory data structure reusing data in iterations. Moreover, it can automatically handle the parallelization and distribute execution of programs on a large cluster. The experiment results show that MatrixMap is 12 times faster than Spark.

A Maturity Analysis of Big Data Technologies

Big Data OverviewDriven by the need to generate business value, the enterprise has started to adopt Big Data as a solution, migrating from the classical databases and data stores which lack the flexibility and are not optimized enough [1].The changes in the environment make big data analytics attractive to all types of organizations, while the market conditions make it practical. The combination of simplified models for development, commoditization, a wider palette of data management tools, and low-cost utility computing has effectively lowered the barrier to entry. [2]. The concept addresses large volumes of complex data, rapid growing data sets that may come from different autonomous sources.In recent approaches, Big Data is characterized by principles known as the 4VVolume, Variety, Velocity and Veracity [3]. There are opinions about accepting other principles as Big Data characteristics, such as Value.Each day more businesses realize that Big Data is relevant as the applications generate large volumes of data generated automatically, from different data sources, centralized or autonomous. As traditional databases hit limitations when the need of analyzing this data, dedicated solutions must be considered.Important BigData Solutions:* Apache HBase/Hadoop is based on Google's BigTable distributed storage system, which runs on top of Hadoop as a distributed and scalable big data store. This means that HBase can leverage the distributed processing paradigm of the Hadoop Distributed File System (HDFS) and benefit from Hadoop's MapReduce programming model. It combines the scalability of Hadoop with real-time data access as a key/value store and deep analytic capabilities of Map Reduce [4].HBase allows to query for individual records as well as derive aggregate analytic reports across a massive amount of data. It can host large tables with billions of rows, millions of columns and run across a cluster of commodity hardware. HBase is composed of three types of servers in a master slave type of architecture. Region servers are responsible to serve data for reads and writes. When accessing data, clients communicate with HBase Region Servers directly. Region assignment, DDL (create, delete tables) operations are handled by the HBase Master process.* Apache Cassandra is a distributed database used for the administration and management of large amounts of structured data across multiple servers, while providing highly available service and no single point of failure. It provides features such as continuous availability, linear scale performance, data distribution across multiple data centers and cloud availability zones. Cassandra inherits its data architecture from Google's BigTable and it borrows its distribution mechanisms from Amazon's Dynamo.The nodes in a Cassandra cluster are completely symmetrical, all having identical responsibilities. Cassandra also employs consistent hashing to partition and replicate data. it has the capability of handling large amounts of data and thousands of concurrent users or operations per second across multiple data centers.Cassandra has a hierarchy of caching mechanisms and carefully orchestrated disk I/O which ensures speed and data safety. Write operations are sent first to a persistent commit log (ensuring a durable write), then to a write-back cache called a memTable; when the memTable fills, it is flushed to a sorted string table - SSTable - on disk. A Cassandra cluster is organized as a ring, and it uses a partitioning strategy to distribute data evenly.* Redis represents an in-memory data structure store used as a database, cache and message broker. It supports data structures such as strings, hashes, lists, sets, sorted sets with range queries, bitmaps, hyperlogs and geospatial indexes with radius queries. Redis stores all data in RAM, allowing lightning fast reads and writes. It runs extremely efficiently in memory and handles high-velocity data, needing simple standard servers to deliver millions of operations per second with submillisecond latency. …

A Heuristic Rule Based Approximate Frequent Itemset Mining Algorithm

In this paper, we focus on the problem of mining the approximate frequent itemsets. To improve the performance, we employ a sampling method, in which a heuristic rule is used to dynamically determine the sampling rate. Two parameters are introduced to implement the rule. Also, we maintain the data synopsis in an in-memory data structure named SFIHtree to speed up the runtime. Our proposed algorithm SFIH can be efficiently performed over this tree. We conducted extensive experiments and showed that the mining performance can be improved significantly with a high accuracy when we used reasonable parameters.

REWIND

Recent non-volatile memory (NVM) technologies, such as PCM, STT-MRAM and ReRAM, can act as both main memory and storage. This has led to research into NVM programming models, where persistent data structures remain in memory and are accessed directly through CPU loads and stores. Existing mechanisms for transactional updates are not appropriate in such a setting as they are optimized for block-based storage. We present REWIND, a user-mode library approach to managing transactional updates directly from user code written in an imperative general-purpose language. REWIND relies on a custom persistent in-memory data structure for the log that supports recoverable operations on itself. The scheme also employs a combination of non-temporal updates, persistent memory fences, and lightweight logging. Experimental results on synthetic transactional workloads and TPC-C show the overhead of REWIND compared to its non-recoverable equivalent to be within a factor of only 1.5 and 1.39 respectively. Moreover, REWIND outperforms state-of-the-art approaches for data structure recoverability as well as general purpose and NVM-aware DBMS-based recovery schemes by up to two orders of magnitude.

Mobiiscape: Middleware support for scalable mobility pattern monitoring of moving objects in a large-scale city

With the explosive proliferation of mobile devices such as smartphones, tablets, and sensor nodes, location-based services are getting even more attention than before, considered as one of the killer applications in the upcoming mobile computing era. Developing location-based services necessarily requires an effective and scalable location data processing technology. In this paper, we present Mobiiscape, a novel location monitoring system that collectively monitors mobility patterns of a large number of moving objects in a large-scale city to support city-wide mobility-aware applications. Mobiiscape provides an SQL-like query language named Moving Object Monitoring Query Language ( MQL) that allows applications to intuitively specify Mobility Pattern Monitoring Queries ( MPQs). Further, Mobiiscape provides a set of scalable location monitoring techniques to efficiently process a large number of MPQs over a large number of location streams. The scalable processing techniques include a (1) Place Border Index, a spatial index for quickly searching for relevant queries upon receiving location streams, (2) Place-Based Window, a spatial-purpose window for efficiently detecting primitive mobility patterns, (3) Shared NFA, a shared query processing technique for efficiently matching complex mobility patterns, and (4) Attribute Pre-matching Bitmap, an in-memory data structure for efficiently filtering out moving objects based on their attributes. We have implemented a Mobiiscape prototype system. Then, we show the usefulness of the system by implementing promising location-based applications based on it such as a ubiquitous taxicab service and a location-based advertising. Also, we demonstrate the performance benefit of the system through extensive evaluation and comparison.

FRASH

In this work, we develop a novel hybrid file system, FRASH, for storage-class memory and NAND Flash. Despite the promising physical characteristics of storage-class memory, its scale is an order of magnitude smaller than the current storage device scale. This fact makes it less than desirable for use as an independent storage device. We carefully analyze in-memory and on-disk file system objects in a log-structured file system, and exploit memory and storage aspects of the storage-class memory to overcome the drawbacks of the current log-structured file system. FRASH provides a hybrid view storage-class memory. It harbors an in-memory data structure as well as a on-disk structure. It provides nonvolatility to key data structures which have been maintained in-memory in a legacy log-structured file system. This approach greatly improves the mount latency and effectively resolves the robustness issue. By maintaining on-disk structure in storage-class memory, FRASH provides byte-addressability to the file system object and metadata for page, and subsequently greatly improves the I/O performance compared to the legacy log-structured approach. While storage-class memory offers byte granularity, it is still far slower than its DRAM counter part. We develop a copy-on-mount technique to overcome the access latency difference between main memory and storage-class memory. Our file system was able to reduce the mount time by 92% and file system I/O performance was increased by 16%.

Execution of distributed queries in the web environment

Database query processing is one of the fundamental issues in data engineering, particularly when more and more databases become distributed over the Internet in the last decade. There have been several approaches to the problem of distributed query processing (DQP), some were successful in the client-server environment but not so in an environment with heterogeneous and autonomous databases. The challenge relies on the nature of the Internet and the Web that pose many difficulties to the problem in hand. In this paper, we propose a model that uses an XML standard and its equivalent in-memory data structure to achieve data sharing, and invokes XML Web services in asynchronous mode to allow local database management systems to efficiently process local data. This approach, like XJoin, allows progress to be made even when data sources experience delays. It builds a layer on top of existing database management systems to minimize the complexity and cost for processing distributed queries that is critical for making such a method practical.

BIRCH: A New Data Clustering Algorithm and Its Applications

Data clustering is an important technique for exploratory data analysis, and has been studied for several years. It has been shown to be useful in many practical domains such as data classification and image processing. Recently, there has been a growing emphasis on exploratory analysis of very large datasets to discover useful patterns and/or correlations among attributes. This is called data mining, and data clustering is regarded as a particular branch. However existing data clustering methods do not adequately address the problem of processing large datasets with a limited amount of resources (e.g., memory and cpu cycles). So as the dataset size increases, they do not scale up well in terms of memory requirement, running time, and result quality. In this paper, an efficient and scalable data clustering method is proposed, based on a new in-memory data structure called CF-tree, which serves as an in-memory summary of the data distribution. We have implemented it in a system called BIRCH (Balanced Iterative Reducing and Clustering using Hierarchies), and studied its performance extensively in terms of memory requirements, running time, clustering quality, stability and scalability; we also compare it with other available methods. Finally, BIRCH is applied to solve two real-life problems: one is building an iterative and interactive pixel classification tool, and the other is generating the initial codebook for image compression.