Arenadata Command Center (ADCC)

Main article: Data management

2021: ADCC version 3.0

Arenadata announced on June 24, 2021 the release of an updated version of the request monitoring system in Arenadata DB (analytical MPP-DBMS based on Greenplum) - Arenadata Command Center (ADCC) 3.0. ADCC allows you to monitor the execution of queries in Arenata DB, evaluate cluster workload, analyze query plans, and perform some other administrative tasks that occur during database operation.

With this service, you can obtain metrics such as query execution parameters, database object locks, temporary query data, spill files, and so on. Prior to the introduction of Command Center, such monitoring could be carried out only by built-in database tools that did not provide the necessary level of detail and up-to-date data, and were not simple and easy to use.

ADCC uses the built-in gpperfmon utility and the gpADCC extension as tools for collecting metrics. The data collected by them is sent to agents - special applications installed on each segment-host, engaged in the transfer of metrics to the ADCC server for subsequent storage in the service database (PostgreSQL) and display in the Web interface of the system.

Prior to version 3.0, the Web server, which was deployed in a single instance, was engaged in both the collection and aggregation of metrics on the server side and the processing of user requests from UI, which negatively affected the scalability and performance of the system as a whole, making it not adapted to high loads.

Compared to previous versions of the system, ADCC 3.0 solved problems of performance and correctness of the user interface, as well as improved scalability of the system. This was achieved through a number of changes at the architecture level, including a horizontally scalable service for processing and aggregating metrics, removing tasks from the Web server to receive metrics and write them to the database, asynchronous processing of metrics through the use of queues and thread pools, filtering metrics not processed by the system, load balancing, etc.

We have developed a system that not only solves the performance and scalability problems of the previous version, but also is capable of handling significantly large amounts of data. By doing so, our customers "database administrators and analysts will be able to track the status of requests in near real-time. said Dmitry Pluzhnikov, director of the Arenata system architecture department.

The architecture of version 3.0 makes ADCC more fault tolerant, accessible, productive, and easy to operate. The system is also horizontally scalable, which allows it to work with clusters of various complexity. In order to facilitate the transition between different versions of the system, a separate utility for migrating its data has been written. Future product development plans include increasing user interface functionality, adding types of metrics, increasing detail on already processed metrics, and dividing the metrics store into cold (with rare access) and hot (with frequent access). In the future, Greenplum 7 support and the ability to monitor several ADB clusters will be added.