A Query, a Minute: Evaluating Performance Isolation in Cloud Databases

Kiefer, Tim; Schön, Hendrik; Habich, Dirk; Lehner, Wolfgang

doi:10.1007/978-3-319-15350-6_11

Cited by 4 publications

(6 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This differs from many benchmarks conducted in database research, where database workload often runs in isolation, while multi-tenant environments are closer to real-world conditions. Similarly, an overview over performance isolation for cloud databases is provided in [Ki15].…”

Section: Related Workmentioning

confidence: 99%

“…While general-purpose operating systems (like Linux) provide a great deal of hardware support, drivers and system abstractions, they have also been identified as a cause of jitter in network bandwidth, disk I/O, or CPU [Ar09; SDQ10; Xu13] when operating software services in cloud environments, where multiple tenants compete for resources. Naturally, this also affects the performance of cloud-hosted database engines [Ki15].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Silentium! Run-Analyse-Eradicate the Noise out of the DB/OS Stack

Mauerer,

Ramsauer,

Lucas

et al. 2021

Preprint

View full text Add to dashboard Cite

When multiple tenants compete for resources, database performance tends to suffer. Yet there are scenarios where guaranteed sub-millisecond latencies are crucial, such as in real-time data processing, IoT devices, or when operating in safety-critical environments. In this paper, we study how to make query latencies deterministic in the face of noise (whether caused by other tenants or unrelated operating system tasks). We perform controlled experiments with an in-memory database engine in a multi-tenant setting, where we successively eradicate noisy interference from within the system software stack, to the point where the engine runs close to bare-metal on the underlying hardware.We show that we can achieve query latencies comparable to the database engine running as the sole tenant, but without noticeably impacting the workload of competing tenants. We discuss these results in the context of ongoing efforts to build custom operating systems for database workloads, and point out that for certain use cases, the margin for improvement is rather narrow. In fact, for scenarios like ours, existing operating systems might just be good enough, provided that they are expertly configured. We then critically discuss these findings in the light of a broader family of database systems (e.g. including disk-based), and how to extend the approach of this paper accordingly.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Silentium! Run-Analyse-Eradicate the Noise out of the DB/OS Stack

Mauerer,

Ramsauer,

Lucas

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…For example, Amazon RDS estimates a one minute I/O suspension on the master DB as the database snapshot is created for replication [16]. Further, DBaaS offerings have been shown to exhibit high variability in query response times for single-instance relational databases [30]. For DBAs, this translates to a trade-off between adding a new replica to achieve better performance and the probability of SLA violations during replica creation.…”

Section: Motivating Examplementioning

confidence: 99%

“…DBaaS providers have automated fundamental database administrative services, however, this in turn has reduced DBA control on many of the performance aspects of the database [30], especially since service providers do not expose details about collocation, physical resources and data placement. For example, limiting replicas to read-only functionality on Amazon RDS prevents the collection of query-level statistical data from these replicas.…”

Section: A Modeling Challengesmentioning

confidence: 99%

Quantifying the Impact of Replication on the Quality-of-Service in Cloud Databases

Osman

Pérez

Casale

2016

2016 IEEE International Conference on Software Quality, Reliability and Security (QRS)

View full text Add to dashboard Cite

Abstract-Cloud databases achieve high availability by automatically replicating data on multiple nodes. However, the overhead caused by the replication process can lead to an increase in the mean and variance of transaction response times, causing unforeseen impacts on the offered quality-of-service (QoS). In this paper, we propose a measurement-driven methodology to predict the impact of replication on Database-as-a-Service (DBaaS) environments. Our methodology uses operational data to parameterize a closed queueing network model of the database cluster together with a Markov model that abstracts the dynamic replication process. Experiments on Amazon RDS show that our methodology predicts response time mean and percentiles with errors of just 1% and 15% respectively, and under operational conditions that are significantly different from the ones used for model parameterization. We show that our modeling approach surpasses standard modeling methods and illustrate the applicability of our methodology for automated DBaaS provisioning.

show abstract

“…Proposed methods require complicated profiling tasks to generate models and specialize to one type of workload (e.g., Relational database [33] or MapReduce [26]). In addition, there is inherent uncertainty in prediction, caused by multitenancy common in the cloud [66,55,17,62,34]. Lastly, users' preferences are complex, involving both completion time [53] and cost [51,82,35,84,38], which have been considered as separate goals [27,42,44], but have not been successfully integrated.…”

Section: Introductionmentioning

confidence: 99%

A Consumer-Centric Market for Database Computation in the Cloud

Wang,

Meliou,

Miklau

2016

Preprint

View full text Add to dashboard Cite

The availability of public computing resources in the cloud has revolutionized data analysis, but requesting cloud resources often involves complex decisions for consumers. Under the current pricing mechanisms, cloud service providers offer several service options and charge consumers based on the resources they use. Before they can decide which cloud resources to request, consumers have to estimate the completion time and cost of their computational tasks for different service options and possibly for different service providers. This estimation is challenging even for expert cloud users.We propose a new market-based framework for pricing computational tasks in the cloud. Our framework introduces an agent between consumers and cloud providers. The agent takes data and computational tasks from users, estimates time and cost for evaluating the tasks, and returns to consumers contracts that specify the price and completion time. Our framework can be applied directly to existing cloud markets without altering the way cloud providers offer and price services. In addition, it simplifies cloud use for consumers by allowing them to compare contracts, rather than choose resources directly. We present design, analytical, and algorithmic contributions focusing on pricing computation contracts, analyzing their properties, and optimizing them in complex workflows. We conduct an experimental evaluation of our market framework over a real-world cloud service and demonstrate empirically that our market ensures three key properties: (a) that consumers benefit from using the market due to competitiveness among agents, (b) that agents have an incentive to price contracts fairly, and (c) that inaccuracies in estimates do not pose a significant risk to agents' profits. Finally, we present a fine-grained pricing mechanism for complex workflows and show that it can increase agent profits by more than an order of magnitude in some cases.

show abstract

A Query, a Minute: Evaluating Performance Isolation in Cloud Databases

Cited by 4 publications

References 8 publications

Silentium! Run-Analyse-Eradicate the Noise out of the DB/OS Stack

Silentium! Run-Analyse-Eradicate the Noise out of the DB/OS Stack

Quantifying the Impact of Replication on the Quality-of-Service in Cloud Databases

A Consumer-Centric Market for Database Computation in the Cloud

Contact Info

Product

Resources

About