arXiv subjects

Tim Kraska

Publications and source records attributed to Tim Kraska.

At least 37 records · Page 2Linked to original sources

arXiv2024

Blueprinting the Cloud: Unifying and Automatically Optimizing Cloud Data Infrastructures with BRAD -- Extended Version

Modern organizations manage their data with a wide variety of specialized cloud database engines (e.g., Aurora, BigQuery, etc.). However, designing and managing such infrastructures is hard. Developers must consider many possible designs with non-obvious performance consequences; moreover, current software abstractions tightly couple applications to specific systems (e.g., with engine-specific clients), making it difficult to change after initial deployment. A better solution would virtualize cloud data management, allowing developers to declaratively specify their workload requirements and rely on automated solutions to design and manage the physical realization. In this paper, we present a technique called blueprint planning that achieves this vision. The key idea is to project data infrastructure design decisions into a unified design space (blueprints). We then systematically search over candidate blueprints using cost-based optimization, leveraging learned models to predict the utility of a blueprint on the workload. We use this technique to build BRAD, the first cloud data virtualization system. BRAD users issue queries to a single SQL interface that can be backed by multiple cloud database services. BRAD automatically selects the most suitable engine for each query, provisions and manages resources to minimize costs, and evolves the infrastructure to adapt to workload shifts. Our evaluation shows that BRAD meet user-defined performance targets and improve cost-savings by 1.6-13x compared to serverless auto-scaling or HTAP systems.

cs.DB↗

arXiv2024

A Declarative System for Optimizing AI Workloads

A long-standing goal of data management systems has been to build systems which can compute quantitative insights over large corpora of unstructured data in a cost-effective manner. Until recently, it was difficult and expensive to extract facts from company documents, data from scientific papers, or metrics from image and video corpora. Today's models can accomplish these tasks with high accuracy. However, a programmer who wants to answer a substantive AI-powered query must orchestrate large numbers of models, prompts, and data operations. For even a single query, the programmer has to make a vast number of decisions such as the choice of model, the right inference method, the most cost-effective inference hardware, the ideal prompt design, and so on. The optimal set of decisions can change as the query changes and as the rapidly-evolving technical landscape shifts. In this paper we present Palimpzest, a system that enables anyone to process AI-powered analytical queries simply by defining them in a declarative language. The system uses its cost optimization framework to implement the query plan with the best trade-offs between runtime, financial cost, and output data quality. We describe the workload of AI-powered analytics tasks, the optimization methods that Palimpzest uses, and the prototype system itself. We evaluate Palimpzest on tasks in Legal Discovery, Real Estate Search, and Medical Schema Matching. We show that even our simple prototype offers a range of appealing plans, including one that is 3.3x faster and 2.9x cheaper than the baseline method, while also offering better data quality. With parallelism enabled, Palimpzest can produce plans with up to a 90.3x speedup at 9.1x lower cost relative to a single-threaded GPT-4 baseline, while obtaining an F1-score within 83.5% of the baseline. These require no additional work by the user.

cs.CL↗

arXiv2024

SEED: Domain-Specific Data Curation With Large Language Models

Data curation tasks that prepare data for analytics are critical for turning data into actionable insights. However, due to the diverse requirements of applications in different domains, generic off-the-shelf tools are typically insufficient. As a result, data scientists often have to develop domain-specific solutions tailored to both the dataset and the task, e.g. writing domain-specific code or training machine learning models on a sufficient number of annotated examples. This process is notoriously difficult and time-consuming. We present SEED, an LLM-as-compiler approach that automatically generates domain-specific data curation solutions via Large Language Models (LLMs). Once the user describes a task, input data, and expected output, the SEED compiler produces a hybrid pipeline that combines LLM querying with more cost-effective alternatives, such as vector-based caching, LLM-generated code, and small models trained on LLM-annotated data. SEED features an optimizer that automatically selects from the four LLM-assisted modules and forms a hybrid execution pipeline that best fits the task at hand. To validate this new, revolutionary approach, we conducted experiments on $9$ datasets spanning over $5$ data curation tasks. In comparison to solutions that use the LLM on every data record, SEED achieves state-of-the-art or comparable few-shot performance, while significantly reducing the number of LLM calls.

cs.DB↗

arXiv2024

PipeRAG: Fast Retrieval-Augmented Generation via Algorithm-System Co-design

Retrieval-augmented generation (RAG) can enhance the generation quality of large language models (LLMs) by incorporating external token databases. However, retrievals from large databases can constitute a substantial portion of the overall generation time, particularly when retrievals are periodically performed to align the retrieved content with the latest states of generation. In this paper, we introduce PipeRAG, a novel algorithm-system co-design approach to reduce generation latency and enhance generation quality. PipeRAG integrates (1) pipeline parallelism to enable concurrent retrieval and generation processes, (2) flexible retrieval intervals to maximize the efficiency of pipeline parallelism, and (3) a performance model to automatically balance retrieval quality and latency based on the generation states and underlying hardware. Our evaluation shows that, by combining the three aforementioned methods, PipeRAG achieves up to 2.6$\times$ speedup in end-to-end generation latency while improving generation quality. These promising results showcase the effectiveness of co-designing algorithms with underlying systems, paving the way for the adoption of PipeRAG in future RAG systems.

cs.CL↗

arXiv2024

Stage: Query Execution Time Prediction in Amazon Redshift

Query performance (e.g., execution time) prediction is a critical component of modern DBMSes. As a pioneering cloud data warehouse, Amazon Redshift relies on an accurate execution time prediction for many downstream tasks, ranging from high-level optimizations, such as automatically creating materialized views, to low-level tasks on the critical path of query execution, such as admission, scheduling, and execution resource control. Unfortunately, many existing execution time prediction techniques, including those used in Redshift, suffer from cold start issues, inaccurate estimation, and are not robust against workload/data changes. In this paper, we propose a novel hierarchical execution time predictor: the Stage predictor. The Stage predictor is designed to leverage the unique characteristics and challenges faced by Redshift. The Stage predictor consists of three model states: an execution time cache, a lightweight local model optimized for a specific DB instance with uncertainty measurement, and a complex global model that is transferable across all instances in Redshift. We design a systematic approach to use these models that best leverages optimality (cache), instance-optimization (local model), and transferable knowledge about Redshift (global model). Experimentally, we show that the Stage predictor makes more accurate and robust predictions while maintaining a practical inference latency and memory overhead. Overall, the Stage predictor can improve the average query execution latency by $20\%$ on these instances compared to the prior query performance predictor in Redshift.

cs.DB↗

arXiv2023

Extract-Transform-Load for Video Streams

Social media, self-driving cars, and traffic cameras produce video streams at large scales and cheap cost. However, storing and querying video at such scales is prohibitively expensive. We propose to treat large-scale video analytics as a data warehousing problem: Video is a format that is easy to produce but needs to be transformed into an application-specific format that is easy to query. Analogously, we define the problem of Video Extract-Transform-Load (V-ETL). V-ETL systems need to reduce the cost of running a user-defined V-ETL job while also giving throughput guarantees to keep up with the rate at which data is produced. We find that no current system sufficiently fulfills both needs and therefore propose Skyscraper, a system tailored to V-ETL. Skyscraper can execute arbitrary video ingestion pipelines and adaptively tunes them to reduce cost at minimal or no quality degradation, e.g., by adjusting sampling rates and resolutions to the ingested content. Skyscraper can hereby be provisioned with cheap on-premises compute and uses a combination of buffering and cloud bursting to deal with peaks in workload caused by expensive processing configurations. In our experiments, we find that Skyscraper significantly reduces the cost of V-ETL ingestion compared to adaptions of current SOTA systems, while at the same time giving robustness guarantees that these systems are lacking.

cs.DB↗

arXiv2023

Parallel External Sorting of ASCII Records Using Learned Models

External sorting is at the core of many operations in large-scale database systems, such as ordering and aggregation queries for large result sets, building indexes, sort-merge joins, duplicate removal, sharding, and record clustering. Unlike in-memory sorting, these algorithms need to work together with the OS and the filesystem to efficiently utilize system resources and minimize disk I/O. In this paper we describe ELSAR: a parallel external sorting algorithm that uses an innovative paradigm based on a learned data distribution model. The algorithm leverages the model to arrange the input records into mutually exclusive, monotonic, and equi-depth partitions that, once sorted, can simply be concatenated to form the output. This method completely eliminates the need for multi-way file merging, which is typically used in external sorting. We present thorough benchmarks for uniform and skewed datasets in various storage media, where we measure the sorting rates, size scalability, and energy efficiency of ELSAR and other sorting algorithms. We observed that ELSAR has up to 1.65x higher sorting rates than the next-best external sort (Nsort) on SSD drives and 5.31x higher than the GNU coreutils' sort utility on Intel Optane non-volatile memory. In addition, ELSAR supersedes the current winner of the SortBenchmark for the most energy-efficient external string sorting algorithm by an impressive margin of 41%. These results reinforce the premise that novel learning-enhanced algorithms can provide remarkable performance benefits over traditional ones.

cs.DB↗

arXiv2022

FactorJoin: A New Cardinality Estimation Framework for Join Queries

Cardinality estimation is one of the most fundamental and challenging problems in query optimization. Neither classical nor learning-based methods yield satisfactory performance when estimating the cardinality of the join queries. They either rely on simplified assumptions leading to ineffective cardinality estimates or build large models to understand the data distributions, leading to long planning times and a lack of generalizability across queries. In this paper, we propose a new framework FactorJoin for estimating join queries. FactorJoin combines the idea behind the classical join-histogram method to efficiently handle joins with the learning-based methods to accurately capture attribute correlation. Specifically, FactorJoin scans every table in a DB and builds single-table conditional distributions during an offline preparation phase. When a join query comes, FactorJoin translates it into a factor graph model over the learned distributions to effectively and efficiently estimate its cardinality. Unlike existing learning-based methods, FactorJoin does not need to de-normalize joins upfront or require executed query workloads to train the model. Since it only relies on single-table statistics, FactorJoin has small space overhead and is extremely easy to train and maintain. In our evaluation, FactorJoin can produce more effective estimates than the previous state-of-the-art learning-based methods, with 40x less estimation latency, 100x smaller model size, and 100x faster training speed at comparable or better accuracy. In addition, FactorJoin can estimate 10,000 sub-plan queries within one second to optimize the query plan, which is very close to the traditional cardinality estimators in commercial DBMS.

cs.DB↗

arXiv2022

ExSample: Efficient Searches on Video Repositories through Adaptive Sampling

Capturing and processing video is increasingly common as cameras become cheaper to deploy. At the same time, rich video understanding methods have progressed greatly in the last decade. As a result, many organizations now have massive repositories of video data, with applications in mapping, navigation, autonomous driving, and other areas. Because state-of-the-art object detection methods are slow and expensive, our ability to process even simple ad-hoc object search queries ('find 100 traffic lights in dashcam video') over this accumulated data lags far behind our ability to collect it. Processing video at reduced sampling rates is a reasonable default strategy for these types of queries, however, the ideal sampling rate is both data and query dependent. We introduce ExSample, a low cost framework for object search over unindexed video that quickly processes search queries by adapting the amount and location of sampled frames to the particular data and query being processed. ExSample prioritizes the processing of frames in a video repository so that processing is focused in portions of video that most likely contain objects of interest. It continually re-prioritizes processing based on feedback from previously processed frames. On large, real-world datasets, ExSample reduces processing time by up to 6x over an efficient random sampling baseline and by several orders of magnitude over state-of-the-art methods that train specialized per-query surrogate models. ExSample is thus a key component in building cost-efficient video data management systems.

cs.DB↗

arXiv2022

LSI: A Learned Secondary Index Structure

Learned index structures have been shown to achieve favorable lookup performance and space consumption compared to their traditional counterparts such as B-trees. However, most learned index studies have focused on the primary indexing setting, where the base data is sorted. In this work, we investigate whether learned indexes sustain their advantage in the secondary indexing setting. We introduce Learned Secondary Index (LSI), a first attempt to use learned indexes for indexing unsorted data. LSI works by building a learned index over a permutation vector, which allows binary search to performed on the unsorted base data using random access. We additionally augment LSI with a fingerprint vector to accelerate equality lookups. We show that LSI achieves comparable lookup performance to state-of-the-art secondary indexes while being up to 6x more space efficient.

cs.DB↗

arXiv2022

The Case for Learned In-Memory Joins

In-memory join is an essential operator in any database engine. It has been extensively investigated in the database literature. In this paper, we study whether exploiting the CDF-based learned models to boost the join performance is practical or not. To the best of our knowledge, we are the first to fill this gap. We investigate the usage of CDF-based partitioning and learned indexes (e.g., Recursive Model Indexes (RMI) and RadixSpline) in the three join categories; indexed nested loop join (INLJ), sort-based joins (SJ) and hash-based joins (HJ). Our study shows that there is a room to improve the performance of INLJ and SJ categories through our proposed optimized learned variants. Our experimental analysis showed that these proposed learned variants of INLJ and SJ consistently outperform the state-of-the-art techniques.

cs.DB↗

arXiv2021

Bounding the Last Mile: Efficient Learned String Indexing

We introduce the RadixStringSpline (RSS) learned index structure for efficiently indexing strings. RSS is a tree of radix splines each indexing a fixed number of bytes. RSS approaches or exceeds the performance of traditional string indexes while using 7-70$\times$ less memory. RSS achieves this by using the minimal string prefix to sufficiently distinguish the data unlike most learned approaches which index the entire string. Additionally, the bounded-error nature of RSS accelerates the last mile search and also enables a memory-efficient hash-table lookup accelerator. We benchmark RSS on several real-world string datasets against ART and HOT. Our experiments suggest this line of research may be promising for future memory-intensive database applications.

cs.DB↗

arXiv2021

Towards Practical Learned Indexing

Latest research proposes to replace existing index structures with learned models. However, current learned indexes tend to have many hyperparameters, often do not provide any error guarantees, and are expensive to build. We introduce Practical Learned Index (PLEX). PLEX only has a single hyperparameter $ε$ (maximum prediction error) and offers a better trade-off between build and lookup time than state-of-the-art approaches. Similar to RadixSpline, PLEX consists of a spline and a (multi-level) radix layer. It first builds a spline satisfying the given $ε$ and then performs an ad-hoc analysis of the distribution of spline points to quickly tune the radix layer.

cs.DB↗

arXiv2021

Defeating duplicates: A re-design of the LearnedSort algorithm

LearnedSort is a novel sorting algorithm that, unlike traditional methods, uses fast ML models to boost the sorting speed. The models learn to estimate the input's distribution and arrange the keys in sorted order by predicting their empirical cumulative distribution function (eCDF) values. LearnedSort has shown outstanding performance compared to state-of-the-art sorting algorithms on several datasets, both synthetic and real. However, given the nature of the eCDF model, its performance is affected in the cases when the input data contains a large number of repeated keys (i.e., duplicates). This work analyzes this scenario in depth and introduces LearnedSort 2.0: a re-design of the algorithm that addresses this issue and enables the algorithm to maintain the leading edge even for high-duplicate inputs. Our extensive benchmarks on a large set of diverse datasets demonstrate that the new design performs at much higher sorting rates than the original version: an average of 4.78x improvement for high-duplicate datasets, and 1.60x for low-duplicate datasets while taking the lead among sorting algorithms.

cs.DS↗

arXiv2021

When Are Learned Models Better Than Hash Functions?

In this work, we aim to study when learned models are better hash functions, particular for hash-maps. We use lightweight piece-wise linear models to replace the hash functions as they have small inference times and are sufficiently general to capture complex distributions. We analyze the learned models in terms of: the model inference time and the number of collisions. Surprisingly, we found that learned models are not much slower to compute than hash functions if optimized correctly. However, it turns out that learned models can only reduce the number of collisions (i.e., the number of times different keys have the same hash value) if the model is able to over-fit to the data; otherwise, it can not be better than an ordinary hash function. Hence, how much better a learned model is in avoiding collisions highly depends on the data and the ability of the model to over-fit. To evaluate the effectiveness of learned models, we used them as hash functions in the bucket chaining and Cuckoo hash tables. For bucket chaining hash table, we found that learned models can achieve 30% smaller sizes and 10% lower probe latency. For Cuckoo hash tables, in some datasets, learned models can increase the ratio of keys stored in their primary locations by around 10%. In summary, we found that learned models can indeed outperform hash functions but only for certain data distributions and with a limited margin.

cs.DB↗

arXiv2021

MISIM: A Neural Code Semantics Similarity System Using the Context-Aware Semantics Structure

Code semantics similarity can be used for many tasks such as code recommendation, automated software defect correction, and clone detection. Yet, the accuracy of such systems has not yet reached a level of general purpose reliability. To help address this, we present Machine Inferred Code Similarity (MISIM), a neural code semantics similarity system consisting of two core components: (i)MISIM uses a novel context-aware semantics structure, which was purpose-built to lift semantics from code syntax; (ii)MISIM uses an extensible neural code similarity scoring algorithm, which can be used for various neural network architectures with learned parameters. We compare MISIM to four state-of-the-art systems, including two additional hand-customized models, over 328K programs consisting of over 18 million lines of code. Our experiments show that MISIM has 8.08% better accuracy (using MAP@R) compared to the next best performing system.

cs.LG↗

arXiv2021

LEA: A Learned Encoding Advisor for Column Stores

Data warehouses organize data in a columnar format to enable faster scans and better compression. Modern systems offer a variety of column encodings that can reduce storage footprint and improve query performance. Selecting a good encoding scheme for a particular column is an optimization problem that depends on the data, the query workload, and the underlying hardware. We introduce Learned Encoding Advisor (LEA), a learned approach to column encoding selection. LEA is trained on synthetic datasets with various distributions on the target system. Once trained, LEA uses sample data and statistics (such as cardinality) from the user's database to predict the optimal column encodings. LEA can optimize for encoded size, query performance, or a combination of the two. Compared to the heuristic-based encoding advisor of a commercial column store on TPC-H, LEA achieves 19% lower query latency while using 26% less space.

cs.DB↗

arXiv2021

TagMe: GPS-Assisted Automatic Object Annotation in Videos

Training high-accuracy object detection models requires large and diverse annotated datasets. However, creating these data-sets is time-consuming and expensive since it relies on human annotators. We design, implement, and evaluate TagMe, a new approach for automatic object annotation in videos that uses GPS data. When the GPS trace of an object is available, TagMe matches the object's motion from GPS trace and the pixels' motions in the video to find the pixels belonging to the object in the video and creates the bounding box annotations of the object. TagMe works using passive data collection and can continuously generate new object annotations from outdoor video streams without any human annotators. We evaluate TagMe on a dataset of 100 video clips. We show TagMe can produce high-quality object annotations in a fully-automatic and low-cost way. Compared with the traditional human-in-the-loop solution, TagMe can produce the same amount of annotations at a much lower cost, e.g., up to 110x.

cs.CV↗