Comparing Scaling Methods for Linux Containers

Adaptive Microservice Scaling for Elastic Applications

“…but not so much on "What to scale?". As we have mentioned in the introduction, the primary reason for this is that most studies take a load balanced multi-tier web application architecture as the model and a monolithic multi-service business logic layer, hosted on a VM or a container, as the object of the scaling decision [2], [5], [20], [21], [22]. In contrast, our work focuses on the emerging research area of microservice based web application scaling.…”

“…One of the main issues is that, despite appearing simple, to set the right threshold requires a deep knowledge of how the application operates [21]. Another key issue is that due to the inherently "reactive" nature of the threshold-based approach and the fact that any scaling action has some non-trivial delay attached to it (time-to-scale or TTS) it is likely that the performance of the application will degrade while the additional resources are being provisioned [22].…”

Docker Container Deployment in Distributed Fog Infrastructures with Checkpoint/Restart

“…The closest work to ours is an auto-scaling method for containers in large-scale systems. Nadgowda et al proposed a method based on CRIU [26] to checkpoint a running container and efficiently replicate it in other nodes [24]. However, this solution only aims at horizontal scalability scenarios and does not reduce the container initialization latency in the standalone application deployment.…”

“…This is particularly true for complex applications such as MySQL which need to read configuration data and to undergo complex initialization procedures during their boot phase. Booting a mysql container on a fast server requires about 10 s before being available to serve end-user commands [24]. This delay may have a significant impact on the performance of the fog applications as the same application is repeatedly launched, created and booted in a PoP.…”

Containerized IoT Solution for Efficient Mobile Vertical Handover

“…LXC is an operating system level virtualization method for running multiple isolated linux systems on a controlled host using a linux kernel. The Advantages of Linux containers [25] as are follow: 1) Portability: Linux containers can run in any environment without changing the functionality of the operating system. The applications running inside a container can also be bundled together and then deployed onto various environments.…”