Wyniki wyszukiwania - BazTech

1

Thespis: Causally-consistent OLTP

Camilleri Carl, Vella Joseph G., Nezval Vitezslav

Annals of Computer Science and Information Systems

|

2021

|

Vol. 25

261--269

EN

Data Consistency defines the validity of a data set according to some set of rules, and different levels of data consistency have been proposed. Causal consistency is the strongest type of consistency possible when data is stored in multiple locations, and fault tolerance is desired. Thespis is a middleware that leverages the Actor model to implement causal consistency over a DBMS, whilst abstracting complexities for application developers behind a REST interface. ThespisTRX is an extension that provides read-only transaction capabilities, whilst ThespisDIIP is another extension that handles distributed integrity invariant preservation. Here, we analyse standard transactional workloads on the relational data model, which is richer than the key-value data model supported by the Thespis interface. We show the applicability of the Thespis approach for this data model by designing new operations for the Thespis interface, which ensure correct execution of such workloads in a convergent, causally consistent distributed environment.

2

Fed-agent – a Transparent ACID-Enabled Transactional Layer for Multidatabase Microservice Architectures

Nikolić Lazar, Dimitrieski Vladimir

Annals of Computer Science and Information Systems

|

2021

|

Vol. 25

489--492

EN

With the recent expansion of specialized databases and departure from the "one size fits all" paradigm, engineers might decide to use multiple databases. Each database holds a representation of a data object but offers transactions and consistency guarantees only locally. Existing solutions either require additional coding or do not provide global ACID transactions. In this paper, we present fed-agent, a transactional layer that provides global consistency and ACID transactions for single data objects within multidatabase systems. It requires no additional coding besides configuration files. We show that fed-agent scales linearly and introduces an overhead small enough for most microservice solutions.

3

Systolic computation for conflict elimination between parallel executed distributed transactions

Smoliński M.

Metody Informatyki Stosowanej

|

2011

|

nr 4

223-229

EN

Serializable history defines execution order for every distributed transaction requested by enterprise information systems, that share many data source i.e. databases. However parallel execution of distributed transactions could cause a conflict. Many conflicts could have a negative effect as a deadlock between executed transactions and stop further processing in associated enterprise informational systems. Prepared rigorous history eliminates conflicts between distributed transactions. All required computations are processed ahead by dedicated hardware unit, a systolic array. Designed systolic algorithm base on simple condition enabling transaction conflict detection. Main features of systolic array are parallel and streaming processing. Designed systolic algorithm that supports transaction conflict detection is also flexible in usage for various distributed environment.

4

Systoliczne wspomaganie obliczeń przy opracowaniu rygorystycznej historii szeregowania transakcji

Smoliński M.

Prace Instytutu Elektrotechniki

|

2010

|

Z. 246

97-109

PL

Historia szeregowania określa kolejność wykonania transakcji zgłoszonych przez systemy biznesowe wykorzystujące współdzielone bazy danych w środowisku rozproszonym. Występowanie konfliktów pomiędzy operacjami transakcji uwzględnionych w rygorystycznej historii szeregowania uniemożliwia równoległe ich wykonanie. W szczególności historia rygorystyczna eliminuje wystąpienie problemu wzajemnej blokady transakcji rozproszonych. Wymagane obliczenia, niezbędne do przygotowania przez opracowanego planistę rygorystycznej historii wykonania transakcji, są wykonywane przez dedykowany układ sprzętowy. Zaprojektowana liniowa tablica systoliczna zapewnia przetwarzanie równoległe i potokowe przy wykonywaniu obliczeń wspomagających opracowanie rygorystycznej historii szeregowania transakcji.

EN

Serializable history defines execution order for every transaction requested by enterprise information systems, that share many databases in distributed environment. When conflict appears between operations, belonging to different transactions from rigorous history, that makes impossible to execute them in parallel. Especially rigorous history eliminates distributed transactions deadlock problem. Required computations, that are necessary to creation of rigorous history, are executed by dedicated hardware unit. Designed linear, systolic array uses parallel and streaming processing when execute computations to create rigorous history.