Isolation Levels (IL) act as correct contracts between applications and database management systems (DBMSs). The complex code logic and concurrent interactions among transactions make it a hard problem to expose violations of various ILs stated by DBMSs. With the recent proliferation of new DBMSs, especially the cloud ones, there is an urgent demand for a general way to detect bugs violating various ILs. The core challenges come from the requirements of: (a) lightweight (verifying without modifying the application logic in workloads and the source code of DBMSs), (b) generality (verifying various ILs), and (c) efficiency (performing efficient verification on a long running workload). To this end, we propose a powerful and practical bug-finding toolLeopard. For lightweight, we propose to infer transaction dependencies based on the time intervals of operations collected from the client-side, without modifying the source code of DBMSs. For generality, based on a thorough analysis of existing concurrency control protocols, we summarize and abstract four mechanisms which can implement ILs in all commercial DBMSs we have investigated. For efficiency, we design atwo-level pipelineto organize and sort massive time intervals in a time and memory conservative way; we propose amechanism-mirrored verificationto simulate the concurrency control protocols implemented in DBMSs for high throughputs; From experimental results,Leopardoutperforms existing methodsCobraandElle. In practice,Leopardhas a superpower to verify various ILs on any workload running on all commercial DBMSs. Moreover, it has successfully discovered 49 bugs that cannot be found by other existing methods.

Data driven business applications in logistics industry often issue prediction queries over relational databases to retrieve the newly generated transaction data for feature computations. In some cases even using slightly outdated data can result in significant inaccuracies in predictions. In another cases, we also observed the accuracy of model prediction is not sensitive to the data freshness. In the setting of primary-backup databases, one may choose to fetch the freshest data from the primary database to ensure model accuracy. However, this may hurt the performance of read-write transactions on the primary, especially when subjected to a high volume of prediction requests. In this work, we propose a Model-Accuracy Aware Service that facilitates a flexible trade-off between model prediction accuracy and primary database performance. This service implements an automated routing strategy aimed at minimizing the impact on the primary database's performance while meeting the requirements of model accuracy. It achieves this by leveraging the maintained database freshness information and the predictive results feedback from the model to learn the relationship between data discrepancy and prediction discrepancy in primary-backup scenarios. We report the experimental results on a real logistics application and also show its effectiveness on a public dataset.

Throughout index tuning, existing index advisors allocate tuning budget equally across all queries in the workload, even though a considerable portion of queries benefit negligible from index tuning, leading to high costs and inefficiency. This paper introduces a novel learning-based index advisor named GITEE, which increases tuning efficiency and effectiveness by intelligently guiding the exploration of the large search space on candidate index. Our solution consists of three components. First, we utilize execution plan and predicate information to accurately estimate the maximum improvement indexing can bring, which serves as preliminary knowledge for reasonable tuning budget allocation. Second, we filter out queries based on the impact of indexing on the individual queries and their influence on others, thereby reducing the number of candidate indexes. Third, we leverage a Monte Carlo Tree Search-based solution, guided by the knowledge, to accelerate the selection of high-quality index configurations within the valuable search space. Extensive experiments across various benchmarks demonstrate that GITEE achieves superior tuning performance compared to state-of-theart heuristic or learning-based index advisors, while reducing tuning overhead by 1-2 orders of magnitude.

Update-intensive workloads are prevalent in contemporary OLTP and AI/ML scenarios. An update operation typically involves deleting the old version of the target record and then inserting a new version. In this work, we demonstrate that an LSM-tree faces two issues when dealing with update-intensive workloads. Firstly, the deleted old versions are not promptly garbage collected until they merge with their new versions during compaction. This may lead to space waste and write amplification. Secondly, it is common for an update operation to modify only a small fraction of a data record, such as one of a hundred attributes. However, state-of-the-art LSM-trees fail to effectively utilize the incremental storage strategy, which involves storing only the updated fraction rather than the entire new version to enhance efficiency. In this paper, we propose two techniques, active and fast garbage collection, and adaptive incremental updating, to address these issues, respectively. Active and fast garbage collection probes the distribution of invalid data versions in an LSM-tree and performs garbage collection in a more promptly manner. Adaptive incremental updating applies different storage modes to the update operation to achieve balanced write and read amplification ratios as much as possible. Based on the techniques, we introduce SylphDB implemented based on the codebase of RocksDB and optimized for update-intensive workloads. Experimental results demonstrated that, compared to traditional LSM-tree based systems, SylphDB can improve the efficiency of garbage collection by 2× and reduce write amplification by 20%.

In China, the number of riders in the on-demand delivery industry has surpassed ten million. Ensuring that these riders earn a decent income can enhance their financial security, reduce poverty, and promote social equity and stability. Due to ease of use, lower-cost maintenance and environmental friendliness, electric bicycles (e-bikes) are the primary mode of transportation for delivery riders. However, these riders frequently encounter depleted batteries due to limited capacity and prolonged charging times, necessitating inconvenient swaps or recharges during deliveries. To address this issue, we propose the e-bike Battery Swap-as-a-Service (eBaaS), an innovative battery-swapping system that leverages an intelligent AIoT network for seamless battery swapping at distributed locations across urban areas. eBaaS integrates edge-cloud collaboration, battery resource allocation, battery anomaly detection, and battery range prediction to minimize downtime and reduce unnecessary mileage. While eBaaS's potential benefits are evident, there has been a lack of robust methods to quantify its impact. Thus, we further developed the eBaaS Impact Evaluation Method (EIEM), the first comprehensive model to address this gap. EIEM analyzes data from approximately 260,000 delivery riders and 5 million riding trajectories. Findings indicate that eBaaS reduces average invalid mileage by 6 km and increases the order volume by an average of over 20% daily per e-bike rider. Meanwhile, the annual electricity savings result in a reduction of 2.74 million kilograms of carbon emissions for 260,000 riders. The eBaaS system is therefore significantly beneficial for environmental conservation and sustainable urban development.

Previous learning-based parameter query optimization (PQO) methods excel in static workloads by precisely selecting optimal plans in a cache with a fixed set of representative plans. However, these methods struggle in dynamic workloads because they cannot predict over dynamically changed plan caches. These queries that fall outside the previously observed query parameter distribution have the risk of reusing suboptimal plans. Unlike traditional PQO methods that learn mappings from parametric query parameters to a fixed set of execution plans, our adaptive PQO framework ( APQO ) takes both query parameters and the plans themselves as model inputs, thereby handling variable numbers of plans in dynamic workloads. By embedding plan representations through representation learning, we pre-train a foundation model offline, enabling APQO to acquire a generalizable plan performance prediction model. Leveraging the foundation model's predictive capability along with a hybrid data augmentation strategy, we train an online calibration model with minimal training data for distribution-shifted new queries, rapidly adapting knowledge for reusing new plans. APQO is natively designed to handle the characteristics of dynamic workloads. Experimental results show that APQO outperforms existing PQO methods in dynamic workloads, achieving a higher cache hit ratio and significantly reducing query latency.

Distributed databases are widely used in various fields, such as financial services and e-commerce. These businesses generally exhibit characteristics of large-scale and rapid growth. However, these business systems often suffer from deadlocks that prevent them from operating normally for extended periods. Traditional deadlock detection methods face challenges in scalability and efficiency, especially as the number of nodes increases. Therefore, deadlock detection has always been a research area in distributed databases. In this paper, we introduce an efficient deadlock detection algorithm called HAWK, leveraging a H ierarchical A pproach based on W or K load modeling. Our algorithm addresses these issues by constructing a dynamic hierarchical detection tree that adapts to transaction patterns, significantly reducing time complexity and communication overhead. HAWK first models the workload and generates a predicted access graph (PAG), transforming the problem of partitioning detection task in the basic hierarchical detection into partition detection zone (DZ) in the PAG by a graph-cutting algorithm. Then, leveraging the properties of strongly connected components (SCCs) and deadlock cycles, the SCC-cut algorithm naturally partitions the system-wide deadlock detection into multiple non-intersecting detection zones, thereby enhancing detection efficiency. We used the greedy SCC-cut algorithm to perform a more fine-grained partitioning of the complex PAG. Finally, by periodically sampling and updating the hierarchical structure, the algorithm remains responsive to dynamic workload variations, ensuring efficient detection. Our approach outperforms both centralized and distributed methods, offering a more efficient and adaptive solution. Extensive experimental results demonstrate the effectiveness of the HAWK algorithm, showing significant reductions in the duration of the deadlock and improved system throughput.

Exploratory data analysis (EDA), coupled with SQL, is essential for data analysts involved in data exploration and analysis. However, data analysts often encounter two primary challenges: (1) the need to craft SQL queries skillfully and (2) the requirement to generate suitable visualization types that enhance the interpretation of query results. Due to its significance, substantial research efforts have been made to explore different approaches to address these challenges, including leveraging large language models (LLMs). However, existing methods fail to meet real-world data exploration requirements primarily due to (1) complex database schema, (2) unclear user intent, (3) limited cross-domain generalization capability, and (4) insufficient end-to-end text-to-visualization capability. This paper presents TiInsight, an automated SQL-based cross-domain exploratory data analysis system. First, we propose a hierarchical data context (i.e., HDC), which leverages LLMs to summarize the contexts related to the database schema, which is crucial for open-world EDA systems to generalize across data domains. Second, the EDA system is divided into four components (i.e., stages): HDC generation, question clarification and decomposition, text-to-SQL generation (i.e., TiSQL), and data visualization (i.e., TiChart). Finally, we implemented an end-to-end EDA system with a user-friendly GUI in the production environment at PingCAP. We have also open-sourced all APIs of TiInsight to facilitate research within the EDA community. Through extensive evaluations by a real-world user study, we demonstrate that TiInsight offers remarkable performance compared to human experts. Additionally, TiSQL achieves an execution accuracy of 86.3% on the Spider dataset when using GPT-4. It also attains an execution accuracy of 60.98% on the Bird test dataset.

High-reliability distributed coordination services have become an indispensable part of modern large-scale distributed systems. Popular coordination services (e.g., ZooKeeper) adopt a single-writer design to provide a centralized service for managing system metadata, including various configuration information and data catalogs, and to provide distributed synchronization functions. With the continuous increase in metadata size and the scale of distributed systems, these coordination services gradually become performance bottlenecks due to their limitations in capacity, read and write performance, and scalability. To bridge the gaps, we propose FDBKeeper, a novel solution that enables scalable coordination services on distributed ACID key-value database systems. Our motivation is that transactional key-value stores (i.e., FoundationDB) meet the demands of performance and scalability required by large-scale distributed systems over coordination service. To leverage these advantages, coordination services can be implemented as an upper layer on top of distributed ACID key-value databases. Our experimental results demonstrate that FDBKeeper significantly outperforms ZooKeeper across key metrics. Additionally, FDBKeeper reduces hardware resource costs on average by 33% in the production environment, resulting in substantial monetary cost savings. We have successfully replaced ZooKeeper with FDBKeeper in the production-grade ClickHouse cluster deployment.

To optimize query parallelism techniques, substantial workloads are required with specific query plans and customized output size for each operator (denoted as cardinality constraint). To this end, a rich body of query-aware database generators (QAG) are proposed. However, the complex data dependencies hidden behind queries make previous QAGs suffer from deficiencies in supporting complex operators and controlling the generation errors. In this paper, we design a new generator Mirage supporting well for complex operators with low error bounds for cardinality constraints. First, Mirage leverages Query Rewriting and Set Transforming Rules to decouple dependencies between key and non-key columns, which could help generate each of them individually. Then, for the non-key columns, Mirage abstracts cardinality constraints of operators as placement requirements within each column's domain, and further models the generation problem as a classic bin packing problem. Finally, for the key columns, Mirage proposes a uniform representation of join cardinality constraints for all types of PK-FK joins and partitions the data according to the matching status between PK and F K columns. Then, it formulates the key population as a Constraint Programming problem, which can be solved by an existing CP Solver. The experiments show that Mirage conquers all previous work in either operator support or generation error.

Functionalities of a database system are co-designed and jointly maintain the database performance. Each function-ality usually has its own metrics to evaluate its state. Previous knobs tuning methods regard the database system as a black box and aim to automatically find the optimal configurations by collecting and observing the overall performance data (e.g., transaction throughput per second) under various configuration knobs. However, if a functionality is not running in the tuning phase, its knobs irrelevant to performance changes can also be tuned by existing tools and potential risks would be introduced. To resolve this problem, we design a database knob tuning framework to support functionality-aware knobs tuning. It uses multitask learning to take the database overall performance as the objective of main learning task, and each function module as a separate learning task. This framework enhances the tuning results through learning the relationships between different tasks, and avoids adjusting irrelevant knobs by perceiving the status of functionalities. We validate its generalizability on OceanBase and PostgreSQL. Experimental results show that better performances were achieved on the overall performance and the metrics of various functionalities.

As real-time analytics become increasingly important, more organizations are deploying Hybrid Transactional/An-alytical Processing (HTAP) systems. The HTAP systems, based on a primary/backup replication architecture, usually support real-time read-only queries on backup nodes for the data recently generated by OLTP applications on the primary node. This work is based on the observation that real-time analytical applications often require access to only a fraction of the latest modifications from OLTP applications. However, the state-of-the-art parallel log replay approaches treat all replicated transaction logs equally and replay the entire transaction logs with the same priority which does not take consideration into the OLAP query access pattern. This design can result in increased response latency for real-time applications. This paper presents AETS, an Adaptive Epoch-based Two-Stage log replay framework that implements epoch-based log replay and table group transaction commit. Simultaneously, AETS also takes full account of the table access priority in real-time HTAP workload log replay. It aims to make the data required by analytical queries visible more quickly. Furthermore, AETS includes a two-phase parallel log replay algorithm called TPLR, which achieves lower overhead compared to state-of-the-art algorithms through careful design. We also offer an adaptive fine-grained thread resource allocation method that considers changes in table access patterns over time under thread resource constraints. Our experimental results show that AETS significantly reduces visibility delay for real-time queries. And the results also show that AETS achieves significant replay throughput improvement.

Generating synthetic workloads is essential and critical to the performance evaluation of database systems. When benchmarking database performance for a specific application, the similarity between synthetic workloads and real application workloads determines the credibility of the evaluation results. However, it meets a great challenge to catch workload characteristics for a target online transaction processing (OLTP) application considering the complexity of transaction executions. To address this problem, we propose a workload duplicator (Lauca) that can generate synthetic workloads with highly similar performance metrics compared to a specific application on both centralized and distributed databases. By carefully studying the application-driven workload generation problem, we presentTransaction Logic,Data Access DistributionandPartition Access Distributionto characterize runtime workloads and propose novel generation algorithms to guarantee the high fidelity of synthetic workloads. To the best of our knowledge, Lauca is the first application-driven transactional workload generator. We conduct extensive experiments based on TPC-C, SmallBank and YCSB on both centralized and distributed databases. The experimental results show that Lauca consistently generates high-quality synthetic workloads.

The Database Autonomy Service (DAS) is a platform designed to assist database administrators in managing a large number of database instances within major internet companies. One of the key tasks in DAS is to find missing indexes to improve the slow query execution. In Meituan, a vast array of business lines deploy tens of thousands of MySQL database instances. Consequently, a great number of human-generated index cases are accumulated in the DAS platform. This motivates us to build a data-driven index recommendation system, referred to as idxLearner, which can learn index creation knowledge from human-generated index cases. In this demonstration, users can interact with idxLearner by choosing source databases to construct the training data, training the recommendation model, inputting slow queries for various target databases, and observing the recommended indexes and their evaluation results.

Distributed shared buffer (DSB) is a well-known solution to support multi-master database systems. DSB involves controlling access to shared data among different nodes through a distributed shared buffer and lock-based cache coherence protocols. Existing DSB implementations resolve conflicts at the page level and lack the flexibility required by modern cloud database systems. Authors present HyBuffer which mixes rows and pages in the distributed shared buffer. It enables multiple masters to independently modify different rows on the same page, thereby enhancing concurrency and performance. HyBuffer adopts the hybrid design to track the location information of cached data. A centralized node maintains page locations indicating which masters have cached which pages, and the record locations of a cached page are maintained by the master which has cached this page. This design eliminates the need to redistribute metadata when dynamically adding or removing masters. Experimental results demonstrate that HyBuffer achieves better scalability and performance compared to existing approaches.

To support efficiently and continuously loading large scale historical data into a distributed data management system (DDMS), it needs to balance the bulk workload across machines. The fundamental problem is to estimate the time used to merge currently loaded data (defined as incremental data) into previously loaded data (defined as baseline data) for each partition, referred to as partition merge. In this work, we present a learning-based framework, referred to as LeaBalancer, to balance the merge loads across cluster nodes. In the situation where the system is scheduled to have regular bulk loading tasks, LeaBalancer can learn to predict the partition merge time from the merge logs generated by previous bulk loadings. Nevertheless, it is still difficult to balance the bulk workload only using a single plan phase because of inaccurate merge time prediction or other in-progress heavy workloads during the bulk loading. To resolve this problem, we design a multi-round balancing strategy, and at the beginning of each round LeaBalancer carefully chooses partitions for migration according to the remaining merge loads in each node. Experimental results show that LeaBalancer can adaptively perform load balance under various settings.

Due to the variability of IT applications, the back-end databases usually run the mixed OLTP workload, which comprises a variety of transactions. Some of these transactions are high-conflict and others are low-conflict. Furthermore, high-conflict transactions may contend on different groups of data stored in the database. Without precise admission control, too many transactions with conflict on the same group of records are simultaneously executed by the OLTP engine, and this will lead to the well-known problem of data-contention thrashing. Under mixed OLTP workloads, conflicting transactions would be blocked for a long time or rolled back finally, and other transactions have not enough opportunity to be processed.To achieve the optimal performance for each kind of transaction, we design a discriminative admission control mechanism for shared-everything database, referred to as DAC. DAC can quickly identify and classify high-conflict transactions according to the set of records they try to access, which is defined as a conflict zone. DAC makes admission control over OLTP transactions with the conflict zone as the granularity. By adaptively adjusting the transaction concurrency level for each zone, transaction blocking and waiting among the same kind of high-conflict transactions can be alleviated. Furthermore, thread resources are released to make the execution of low-conflict transactions less affected. We evaluate DAC using a main-memory database prototype and a classical disk-based database system. Experimental results demonstrate that DAC can help OLTP engine significantly improve the performance under mixed OLTP workloads.

Optimistic concurrency control (OCC) protocols validate whether a transaction has conflicts with other concurrent transactions after this transaction completes its execution. In this work, we demonstrate that the validation phase has a great influence on the performance of modern in-memory database systems, especially under heterogeneous workloads. The cost of validating operations in a transaction is determined by two main factors. The first factor is the operation type. An OCC protocol would take much less cost on validating a single-record read operation than validating a key-range scan operation. The second factor is the workload type. Existing schemes in OCC variants for validating key-range scan perform differently under various workloads. Although various validation schemes share the same goal of guaranteeing a transaction schedule to be serializable, there are remarkable differences between the costs they introduced. These observations motivate us to design an optimistic concurrency control which can choose a low-cost validation scheme at runtime, referred to as adaptive optimistic concurrency control (AOCC). First, at transaction-level granularity, AOCC can assign a validation method to a transaction according to the features of its operations. Furthermore, for each operation in a transaction, the validation method is selected according to not only the number of accessed records but also the instant characteristics of workloads. Experimental results show that AOCC has good performance and scalability under heterogeneous workloads mixed with point accesses and predicate queries.

This work addresses the need for efficient key-range validation for a composite OLTP and bulk processing workload characterized by modern enterprise applications. In-memory database system (IMDB), mostly adopting the optimistic concurrency control (OCC) mechanism, performs well if the contention of conventional OLTP workloads is low and each transaction only contains point read/write query with primary key. In this work we present the performance problem of IMDBs under mixed OLTP and bulk processing workloads. The reason is that existing OCC protocols take expensive cost to generate a serializable schedule if the OLTP workload contains bulk processing operations with key-range scan. To this end, we develop an efficient and scalable range optimistic concurrency control (ROCC) which uses logical ranges to track the potential conflicting transactions and to reduce the number of transactions to be validated. At the read phase, a transaction keeps a set of predicates to remember the version and precise scope of scanned ranges, which eliminates the cost of maintaining scanned records. Before entering the validation phase, if the transaction intends to update records in the logical range, it needs to register to the corresponding lock-free list implemented by a circular array. Finally, ROCC filters out unrelated transactions and validates the bulk operation at range level. Experimental results show that ROCC has good performance and scalability under heterogeneous workloads mixed with point access and bulk processing.

Log replication is a key component in highly available database systems. In order to guarantee data consistency and reliability, it is common for modern database systems to utilize Paxos protocol, which is responsible for replicating transactional logs from one primary node to multiple backups. However, the Paxos replication needs to store and synchronize some additional metadata, such as committed log sequence number (commit point), to guarantee the consistency of the database. This increases the overhead of storage and network, which would have a negative impact on the throughput in the update intensive work load. In this paper, we present an implementation of log replication and database recovery methods, which adopts the idea of piggybacking, i.e., commit point can be embedded in the commit logs. This practice not only retains virtues of Paxos replication, but also reduces disk and network IO effectively. We implemented and evaluated our approach in a main memory database system. Our experiments show that the piggybacking method can offer 1.3× higher throughput than typical log replication with synchronization mechanism.

Learning to adapt in a new setting is a common challenge to our knowledge and capability. New life would be easier if we actively pursued supervision from the right mentor chosen with our relevant but limited prior knowledge. This variant principle of active learning seems intuitively useful to many domain adaptation problems. In this paper, we substantiate its power for advancing automatic ranking adaptation, which is important in web search since it's prohibitive to gather enough labeled data for every search domain for fully training domain-specific rankers. For the cost-effectiveness, it is expected that only those most informative instances in target domain are collected to annotate while we can still utilize the abundant ranking knowledge in source domain. We propose a unified ranking framework to mutually reinforce the active selection of informative target-domain queries and the appropriate weighting of source training data as related prior knowledge. We select to annotate those target queries whose documents' order most disagrees among the members of a committee built on the mixture of source training data and the already selected target data. Then the replenished labeled set is used to adjust the importance of source queries for enhancing their rank transfer. This procedure iterates until labeling budget exhausts. Based on LETOR3.0 and Yahoo! Learning to Rank Challenge data sets, our approach significantly outperforms the random query annotation commonly used in ranking adaptation and the active rank learner on target-domain data only.

Like traditional supervised and semi-supervised algorithms, learning to rank for information retrieval requires document annotations provided by domain experts. It is costly to annotate training data for different search domains and tasks. We propose to exploit training data annotated for a related domain to learn to rank retrieved documents in the target domain, in which no labeled data is available. We present a simple yet effective approach based on instance-weighting scheme. Our method first estimates the importance of each related-domain document relative to the target domain. Then heuristics are studied to transform the importance of individual documents to the pairwise weights of document pairs, which can be directly incorporated into the popular ranking algorithms. Due to importance weighting, ranking model trained on related domain is highly adaptable to the data of target domain. Ranking adaptation experiments on LETOR3.0 dataset [27] demonstrate that with a fair amount of related-domain training data, our method significantly outperforms the baseline without weighting, and most of time is not significantly worse than an "ideal" model directly trained on target domain.