No abstract

No abstract

Network communication is the slowest component of many operators in distributed parallel databases deployed for large-scale analytics. Whereas considerable work has focused on speeding up databases on modern hardware, communication reduction has received less attention. Existing parallel DBMSs rely on algorithms designed for disks with minor modifications for networks. A more complicated algorithm may burden the CPUs but could avoid redundant transfers of tuples across the network. We introduce track join, a new distributed join algorithm that minimizes network traffic by generating an optimal transfer schedule for each distinct join key. Track join extends the trade-off options between CPU and network. Track join explicitly detects and exploits locality, also allowing for advanced placement of tuples beyond hash partitioning on a single attribute. We propose a novel data placement algorithm based on track join that minimizes the total network cost of multiple joins across different dimensions in an analytical workload. Our evaluation shows that track join outperforms hash join on the most expensive queries of real workloads regarding both network traffic and execution time. Finally, we show that our data placement optimization approach is both robust and effective in minimizing the total network cost of joins in analytical workloads.

Analytics are at the core of many business intelligence tasks. Efficient query execution is facilitated by advanced hardware features, such as multi-core parallelism, shared-nothing low-latency caches, and SIMD vector instructions. Only recently, the SIMD capabilities of mainstream hardware have been augmented with wider vectors and non-contiguous loads termed gathers. While analytical DBMSs minimize the use of indexes in favor of scans based on sequential memory accesses, some data structures remain crucial. The Bloom filter, one such example, is the most efficient structure for filtering tuples based on their existence in a set and its performance is critical when joining tables with vastly different cardinalities. We introduce a vectorized implementation for probing Bloom filters based on gathers that eliminates conditional control flow and is independent of the SIMD length. Our techniques are generic and can be reused for accelerating other database operations. Our evaluation indicates a significant performance improvement over scalar code that can exceed 3X when the Bloom filter is cache-resident.

No abstract

The increasing main-memory capacity has allowed query execution to occur primarily in main memory. Database systems employ compression, not only to fit the data in main memory, but also to address the memory bandwidth bottleneck. Lightweight compression schemes focus on efficiency over compression rate and allow query operators to process the data in compressed form. For instance, dictionary compression keeps the distinct column values in a sorted dictionary and stores the values as index codes with the minimum number of bits. Packing the bits of each code contiguously, namely horizontal bit packing, has been optimized by using SIMD instructions for unpacking and by evaluating predicates in parallel per processor word for selection scans. Interleaving the bits of codes, namely vertical bit packing, provides faster scans, but incurs prohibitive costs for packing and unpacking. Here, we improve packing and unpacking for vertical bit packing using SIMD instructions, achieving more than an order of magnitude speedup. Also, we optimize horizontal bit packing on the latest CPUs and compare all approaches. While no single variant is better in all cases, vertical bit packing offers a good trade-off by combining the fastest scans with comparably fast packing and unpacking.

String processing tasks are common in analytical queries powering business intelligence. Besides substring matching, provided in SQL by the like operator, popular DBMSs also support regular expressions as selective filters. Substring matching can be optimized by using specialized SIMD instructions on mainstream CPUs, reaching the performance of numeric column scans. However, generic regular expressions are harder to evaluate, being dependent on both the DFA size and the irregularity of the input. Here, we optimize matching string columns against regular expressions using SIMD-vectorized code. Our approach avoids accessing the strings in lockstep without branching, to exploit cases when some strings are accepted or rejected early by looking at the first few characters. On common string lengths, our implementation is up to 2X faster than scalar code on a mainstream CPU and up to 5X faster on the Xeon Phi coprocessor, improving regular expression support in DBMSs.

Data partitioning is a critical operation for manipulating large datasets because it subdivides tasks into pieces that are more amenable to efficient processing. It is often the limiting factor in database performance and represents a significant fraction of the overall runtime of large data queries. This article measures the performance and energy of state-of-the-art software partitioners, and describes and evaluates a hardware range partitioner that further improves efficiency.The software implementation is broken into two phases, allowing separate analysis of the partition function computation and data shuffling costs. Although range partitioning is commonly thought to be more expensive than simpler strategies such as hash partitioning, our measurements indicate that careful data movement and optimization of the partition function can allow it to approach the throughput and energy consumption of hash or radix partitioning.For further acceleration, we describe a hardware range partitioner, or HARP, a streaming framework that offers a seamless execution environment for this and other streaming accelerators, and a detailed analysis of a 32nm physical design that matches the throughput of four to eight software threads while consuming just 6.9% of the area and 4.3% of the power of a Xeon core in the same technology generation.