Analytical engines rely on in-memory data caching to avoid storage accesses and provide timely responses by keeping the most frequently accessed data in memory. Purely frequency- and time-based caching decisions, however, are a proxy of the expected query execution speedup only when storage accesses are significantly slower than in-memory query processing. On the other hand, fast storage offers loading times that approach fully in-memory query response times, rendering purely frequency-based statistics incapable of capturing the impact of a caching decision on query execution. For example, caching the input of a frequent query that spends most of its time processing joins is less beneficial than caching a page for a slightly less frequent but scan-heavy query. Thus, existing caching policies waste valuable memory space to cache input data that offer little-to-no acceleration for analytics. This paper proposes HPCache, a buffer management policy that enables fast analytics on high-bandwidth storage by efficiently using the available in-memory space. HPCache caches data based on the speedup potential instead of relying on frequency-based statistics. We show that, with fast storage, the benefit of in-memory caching varies significantly across queries; therefore, we quantify the efficiency of caching decisions and formulate an optimization problem. We implement HPCache in Proteus and show that (i) estimating speedup potential improves memory space utilization, and (ii) simple runtime statistics suffice to infer speedup. We show that HPCache achieves up to a 1.75x speed-up over frequency-based caching policies by caching column proportions and automatically tuning them. Overall, HPCache enables efficient use of the in-memory space for input caching in the presence of fast storage, without requiring workload predictions.

Most modern in-memory online transaction processing (OLTP) engines rely on multi-version concurrency control (MVCC) to provide data consistency guarantees in the presence of conflicting data accesses. MVCC improves concurrency by generating a new version of a record on every write, thus increasing the storage requirements. Existing approaches rely on garbage collection and chain consolidation to reduce the length of version chains and reclaim space by freeing unreachable versions. However, finding unreachable versions requires the traversal of long version chains, which incurs random accesses right into the critical path of transaction execution, hence limiting scalability. This paper introduces OneShotGC, a new multi-version storage design that eliminates version traversal during garbage collection, with minimal discovery and memory management overheads. OneShotGC leverages the temporal correlations across versions to opportunistically cluster them into contiguous memory blocks that can be released in one shot. We implement OneShotGC in Proteus and use YCSB and TPC-C to experimentally evaluate its performance with respect to the state-of-the-art, where we observe an improvement of up to 2x in transactional throughput.

Modern analytical engines rely on Approximate Query Processing (AQP) to provide faster response times than the hardware allows for exact query answering. However, existing AQP methods impose steep performance penalties as workload unpredictability increases. Specifically, offline AQP relies on predictable workloads to create samples that match the queries in a priori to query execution, reducing query response times when queries match the expected workload. As soon as workload predictability diminishes, existing online AQP methods create query-specific samples with little reuse across queries, producing significantly smaller gains in response times. As a result, existing approaches cannot fully exploit the benefits of sampling under increased unpredictability. We analyze sample creation and propose LAQy, a framework for building, expanding, and merging samples to adapt to the changes in workload predicates. We show the main parameters that affect the sample creation time and propose lazy sampling to overcome the unpredictability issues that cause fast-but-specialized samples to be query-specific. We evaluate LAQy by implementing it in an in-memory code-generation-based scale-up analytical engine to show the adaptivity and practicality of our framework in a modern system. LAQy speeds up online sampling processing as a function of sample reuse ranging from practically zero to full online sampling time and from 2.5x to 19.3x in a simulated exploratory workload.

Analytical engines rely on in-memory caching to avoid disk accesses and provide timely responses by keeping the most frequently accessed data in memory. Purely frequency- & time-based caching decisions, however, are a proxy of the expected query execution speedup only when disk accesses are significantly slower than in-memory query processing. On the other hand, fast storage offers loading times that approach or even outperform fully in-memory query execution response times, rendering purely frequency-based statistics incapable of capturing impact of a caching decision on query execution. For example, caching the input of a frequent query that spends most of its time processing joins is less beneficial than caching a page for a slightly less frequent but scan-heavy query. As a result, existing caching policies waste valuable memory space to cache input data that offer little-to-no acceleration for analytics.

Stream Processing Engines (SPEs) have recently begun utilizing heterogeneous coprocessors (e.g., GPUs) to meet the velocity requirements of modern real-time applications. The massive parallelism and high memory bandwidth of GPUs can significantly increase processing throughput in data-intensive streaming scenarios, such as windowed aggregations. However, previous research only focused on the overall architecture of hybrid CPU-GPU streaming systems and the need for efficient in-GPU window operators was overshadowed by the limited interconnect bandwidth.

Modern Hybrid Transactional/Analytical Processing (HTAP) systems use an integrated data processing engine that performs analytics on fresh data, which are ingested from a transactional engine. HTAP systems typically consider data freshness at design time, and are optimized for a fixed range of freshness requirements, addressed at a performance cost for either OLTP or OLAP. The data freshness and the performance requirements of both engines, however, may vary with the workload. We approach HTAP as a scheduling problem, addressed at runtime through elastic resource management. We model an HTAP system as a set of three individual engines: an OLTP, an OLAP and a Resource and Data Exchange (RDE) engine. We devise a scheduling algorithm which traverses the HTAP design spectrum through elastic resource management, to meet the workload data freshness requirements. We propose an in-memory system design which is non-intrusive to the current state-of-art OLTP and OLAP engines, and we use it to evaluate the performance of our approach. Our evaluation shows that the performance benefit of our system for OLAP queries increases over time, reaching up to 50% compared to static schedules for 100 query sequences, while maintaining a small, and controlled, drop in the OLTP throughput.

Traditionally, analytical database engines have used task parallelism provided by modern multisocket multicore CPUs for scaling query execution. Over the past few years, GPUs have started gaining traction as accelerators for processing analytical queries due to their massively data-parallel nature and high memory bandwidth. Recent work on designing join algorithms for CPUs has shown that carefully tuned join implementations that exploit underlying hardware can outperform naive, hardware-oblivious counterparts and provide excellent performance on modern multicore servers. However, there has been no such systematic analysis of hardware-conscious join algorithms for GPUs that systematically explores the dimensions of partitioning (partitioned versus non-partitioned joins), data location (data fitting and not fitting in GPU device memory), and access pattern (skewed versus uniform). In this paper, we present the design and implementation of a family of novel, partitioning-based GPU-join algorithms that are tuned to exploit various GPU hardware characteristics for working around the two main limitations of GPUs–limited memory capacity and slow PCIe interface. Using a thorough evaluation, we show that: i) hardware-consciousness plays a key role in GPU joins similar to CPU joins and our join algorithms can process 1 Billion tuples/second even if no data is GPU resident, ii) radix partitioning-based GPU joins that are tuned to exploit GPU hardware can substantially outperform non-partitioned hash joins, iii) hardware-conscious GPU joins can effectively overcome GPU limitations and match, or even outperform, state-of-the-art CPU joins.

Modern server hardware is increasingly heterogeneous as hardware accelerators, such as GPUs, are used together with multicore CPUs to meet the computational demands of modern data analytics work-loads. Unfortunately, query parallelization techniques used by analytical database engines are designed for homogeneous multicore servers, where query plans are parallelized across CPUs to process data stored in cache coherent shared memory. Thus, these techniques are unable to fully exploit available heterogeneous hardware, where one needs to exploit task-parallelism of CPUs and data-parallelism of GPUs for processing data stored in a deep, non-cache-coherent memory hierarchy with widely varying access latencies and bandwidth. In this paper, we introduce HetExchange-a parallel query execution framework that encapsulates the heterogeneous parallelism of modern multi-CPU-multi-GPU servers and enables the parallelization of (pre-)existing sequential relational operators. In contrast to the interpreted nature of traditional Exchange, HetExchange is designed to be used in conjunction with JIT compiled engines in order to allow a tight integration with the proposed operators and generation of efficient code for heterogeneous hardware. We validate the applicability and efficiency of our design by building a prototype that can operate over both CPUs and GPUs, and enables its operators to be parallelism- and data-location-agnostic. In doing so, we show that efficiently exploiting CPU-GPU parallelism can provide 2.8x and 6.4x improvement in performance compared to state-of-the-art CPU-based and GPU-based DBMS.