The two areas of online transaction processing (OLTP) and analytical (OLAP) present different challenges for database architectures. Currently, customers with high rates mission-critical transactions have split their data into separate systems, one OLTP so-called warehouse OLAP. While allowing decent rates, this separation has many disadvantages including freshness issues due to the delay caused by only periodically initiating Extract Transform Load-data staging excessive resource...

Finding a good join order is crucial for query performance. In this paper, we introduce the Join Order Benchmark (JOB) and experimentally revisit main components in classic optimizer architecture using complex, real-world data set realistic multi-join queries. We investigate quality of industrial-strength cardinality estimators find that all routinely produce large errors. further show while estimates are essential finding order, performance unsatisfactory if engine relies too heavily on...

Main memory capacities have grown up to a point where most databases fit into RAM. For main-memory database systems, index structure performance is critical bottleneck. Traditional in-memory data structures like balanced binary search trees are not efficient on modern hardware, because they do optimally utilize on-CPU caches. Hash tables, also often used for indexes, fast but only support queries. To overcome these shortcomings, we present ART, an adaptive radix tree (trie) indexing in main...

Advances in virtualization technology are enabling the creation of resource pools servers that permit multiple application workloads to share each server pool. Understanding nature enterprise is crucial properly designing and provisioning current future services such pools. This paper considers issues workload analysis, performance modeling, capacity planning. Our goal automate efficient use when hosting large numbers services. We a trace based approach for management relies on i)...

With modern computer architecture evolving, two problems conspire against the state-of-the-art approaches in parallel query execution: (i) to take advantage of many-cores, all work must be distributed evenly among (soon) hundreds threads order achieve good speedup, yet (ii) dividing is difficult even with accurate data statistics due complexity out-of-order cores. As a result, existing for plan-driven parallelism run into load balancing and context-switching bottlenecks, therefore no longer...

Multi-Version Concurrency Control (MVCC) is a widely employed concurrency control mechanism, as it allows for execution modes where readers never block writers. However, most systems implement only snapshot isolation (SI) instead of full serializability. Adding serializability guarantees to existing SI implementations tends be prohibitively expensive.

In the implementation of hosted business services, multiple tenants are often consolidated into same database to reduce total cost ownership. Common practice is map single-tenant logical schemas in application one multi-tenant physical schema database. Such mappings challenging create because enterprise applications allow extend base schema, e.g., for vertical industries or geographic regions. Assuming workload stays within bounds, fundamental limitation on scalability this approach number...

Two emerging hardware trends will dominate the database system technology in near future: increasing main memory capacities of several TB per server and massively parallel multi-core processing. Many algorithmic control techniques current were devised for disk-based systems where I/O dominated performance. In this work we take a new look at well-known sort-merge join which, so far, has not been focus research scalable data processing as it was deemed inferior to hash joins. We devise suite...

We describe a new deep learning approach to cardinality estimation. MSCN is multi-set convolutional network, tailored representing relational query plans, that employs set semantics capture features and true cardinalities. builds on sampling-based estimation, addressing its weaknesses when no sampled tuples qualify predicate, in capturing join-crossing correlations. Our evaluation of using real-world dataset shows significantly enhances the quality which core problem optimization.

This work aims at reducing the main-memory footprint in high performance hybrid OLTP & OLAP databases, while retaining query and transactional throughput. For this purpose, an innovative compressed columnar storage format for cold data, called Data Blocks is introduced. further incorporate a new light-weight index structure Positional SMA that narrows scan ranges within even if entire block cannot be ruled out. To achieve highest performance, compression schemes of are very light-weight,...

While standardized and widely used benchmarks address either operational or real-time Business Intelligence (BI) workloads, the lack of a hybrid benchmark led us to definition new, complex, mixed workload benchmark, called CH-benCHmark. This bridges gap between established single-workload suites TPC-C for OLTP TPC-H OLAP, executes complex workload: transactional based on order entry processing corresponding TPC-H-equivalent OLAP query suite run in parallel same tables single database system....

MapReduce has emerged as a popular tool for distributed and scalable processing of massive data sets is being used increasingly in e-science applications. Unfortunately, the performance systems strongly depends on an even distribution while scientific are often highly skewed. The resulting load imbalance, which raises time, amplified by high runtime complexity reducer tasks. An adaptive balancing strategy required appropriate skew handling. In this paper, we address problem estimating cost...

Recent research has shown that learned models can outperform state-of-the-art index structures in size and lookup performance. While this is a very promising result, existing are often cumbersome to implement slow build. In fact, most approaches we aware of require multiple training passes over the data.

Recent advancements in learned index structures propose replacing existing structures, like B-Trees, with approximate models. In this work, we present a unified benchmark that compares well-tuned implementations of three against several state-of-the-art "traditional" baselines. Using four real-world datasets, demonstrate can indeed outperform non-learned indexes read-only in-memory workloads over dense array. We investigate the impact caching, pipelining, dataset size, and key size. study...

Non-volatile memory (NVM) is a new storage technology that combines the performance and byte addressability of DRAM with persistence traditional devices like flash (SSD). While these properties make NVM highly promising, it not yet clear how to best integrate into layer modern database systems. Two system designs have been proposed. The first use exclusively, i.e., store all data index structures on it. However, because has higher latency than DRAM, this design can be less efficient...

Graph analytics on social networks, Web data, and communication networks has been widely used in a plethora of applications. Many graph algorithms are based breadth-first search (BFS) traversal, which is not only time-consuming for large datasets but also involves much redundant computation when executed multiple times from different start vertices. In this paper, we propose Multi-Source BFS (MS-BFS), an algorithm that designed to run concurrent BFSs over the same single CPU core while...

So far, transactional memory-although a promising technique-suffered from the absence of an efficient hardware implementation. The upcoming Haswell microarchitecture Intel introduces memory (HTM) in mainstream CPUs. HTM allows for concurrent, atomic operations, which is also highly desirable context databases. On other hand has several limitations that, general, prevent one-to-one mapping database transactions to transactions. In this work we devise building blocks that can be used exploit...

Spatial data is pervasive. Large amount of spatial produced every day from GPS-enabled devices such as cell phones, cars, sensors, and various consumer based applications Uber, location-tagged posts in Facebook, In-stagram, Snapchat, etc. This growth coupled with the fact that queries, analytical or transactional, can be computationally extensive has attracted enormous interest research community to develop systems efficiently process analyze this data. In recent years a lot analytics have...

The performance of transactional database systems is critically dependent on the efficient synchronization in-memory data structures. traditional approach, fine-grained locking, does not scale modern hardware. Lock-free structures, in contrast, very well but are extremely difficult to implement and often require additional indirections. In this work, we argue for a middle ground, i.e., protocols that use only sparingly. We synchronize Adaptive Radix Tree (ART) using two such protocols,...

In this paper, we present a context framework that facilitates the development and deployment of context-aware adaptable Web services. services are provided with information about clients may be utilized to provide personalized behavior. Context is extensible new types at any time without changes underlying infrastructure. processing done by services, plugins, or plugins pre- post-process service messages based on available contextinformation. Both essential for automatic adaption necessity...