China
Africa
Explore chapters and articles related to this topic
Storage and databases for big data
Published in Jun Deng, Lei Xing, Big Data in Radiation Oncology, 2019
Tomas Skripcak, Uwe Just, Ida Schönfeld, Esther G.C. Troost, Mechthild Krause
Relational databases are known for strictly following the atomicity, consistency, isolation, durability (ACID) properties for database transactions and they act as ACID-compliant transaction managers. This prevents concurrent users from working with data in a conflicting way; however, this also comes with a performance penalty and limits the system-scaling ability. The approaches for scaling the database resources fall into two categories: vertical scaling (scale up) and horizontal scaling (scale out). This analogy originates from the typical positioning of hardware in a rack within a data center. Scaling up refers to the addition of new resources (e.g., storage space, memory) to the existing server within one rack, and scaling out refers to the addition of new server nodes to neighboring racks in the server room. Conventional relational databases normally scale up well, but their ability to scale out is limited. Moreover, scaling out relational databases often means losing ACID compliance, which is their biggest advantage in transactional workload scenarios.
The Design and Implementation of the Remote Control and Visualization System for Pulse Application Experiments
Published in Fusion Science and Technology, 2022
Yu Gu, Mengqi Fan, Tengfei Cao, Yuanzhe Zhao, Xiaodan Zhang
According to the theory of computation, DFA is an automaton capable of state transition. Given the state of the automaton and a character belonging to the alphabet Σ of the automaton, it can shift to the next state (which can be the previous state) according to the pre-given transition function.7 Therefore, it is appropriate to apply DFA to the system design of RCVS because DFA can ensure the uniformity of the operating state of the experimental equipment. The problem of concurrency between remote users can be solved by using the four features of database transactions: Atomicity, Consistency, Isolation, Durability (ACID). This is called the ACID principle.8 The processing convention of the D2L algorithm is as follows: DFA first obtains the status of the channel number that needs to be operated through the device model and puts its status into the field named “status” of the DEVICE_CH table in the database. If the status is “IDLE,” the channel is available; otherwise, the backoff algorithm is executed to put the process into the state of waiting. If the channel is available, the user initiates a database transaction and adds an exclusive lock (X lock) to the “status” field of the table to be operated on. At this point, a channel state of the device is locked through the 2PL protocol. After that, DFA will change the corresponding device channel state to “READY.” The user performs the trigger operation, and DFA converts the device channel state to “WORKING” and updates the database field “status” to be the same. At this point, other users can obtain the current device channel state through DFA.After the task of data acquisition is completed, the program will unlock the locked tables in the database. At this point, the 2PL protocol is over, and DFA then transits the device state to “IDLE.”