Paper | Title | Page |
---|---|---|
TUPPC023 | MeerKAT Poster and Demo Control and Monitoring Highlights | 594 |
|
||
The 64-dish MeerKAT Karoo Array Telescope, currently under development, will become the largest and most sensitive radio telescope in the Southern Hemisphere until the Square Kilometre Array (SKA) is completed around 2024. MeerKAT will ultimately become an integral part of the SKA. The MeerKAT project will build on the techniques and experience acquired during the development of KAT-7, a 7-dish engineering prototype that has already proved its worth in practical use, operating 24/7 to deliver useful science data in the Karoo. Much of the MeerKAT development will centre on further refinement and scaling of the technology, using lessons learned from KAT-7. The poster session will present the proposed MeerKAT CAM (Control & Monitoring) architecture and highlight the solutions we are exploring for system monitoring, control and scheduling, data archiving and retrieval, and human interaction with the system. We will supplement the poster session with a live demonstration of the present KAT-7 CAM system. This will include a live video feed from the site as well as the use of the current GUI to generate and display the flow of events and data in a typical observation. | ||
![]() |
Poster TUPPC023 [0.471 MB] | |
THCOBA06 | Virtualization and Deployment Management for the KAT-7 / MeerKAT Control and Monitoring System | 1422 |
|
||
Funding: National Research Foundation (NRF) of South Africa To facilitate efficient deployment and management of the Control and Monitoring software of the South African 7-dish Karoo Array Telescope (KAT-7) and the forthcoming Square Kilometer Array (SKA) precursor, the 64-dish MeerKAT Telescope, server virtualization and automated deployment using a host configuration database is used. The advantages of virtualization is well known; adding automated deployment from a configuration database, additional advantages accrue: Server configuration becomes deterministic, development and deployment environments match more closely, system configuration can easily be version controlled and systems can easily be rebuilt when hardware fails. We chose the Debian GNU/Linux based Proxmox VE hypervisor using the OpenVZ single kernel container virtualization method along with Fabric (a Python ssh automation library) based deployment automation and a custom configuration database. This paper presents the rationale behind these choices, our current implementation and our experience with it, and a performance evalution of OpenVZ and KVM. Tests include a comparison of application specific networking performance over 10GbE using several network configurations. |
||
![]() |
Slides THCOBA06 [5.044 MB] | |