Containers are not a new technology. They emerge in the late seventies, but they get used intensive in the past decade in various systems and companies to achieve high resource utilization, efficient task-packing, process-level isolation. Those properties of containers allow large internet companies like Google, Amazon, and Facebook to satisfy all their applications and big data workloads. These companies can run a dozen tasks at scale, to satisfy all their operational needs. On the other hand, developers can create more efficient software and run it everywhere with the same behavior, almost without any additional work. With emerge of automatic cluster management and cluster schedulers, these containers get used even more on a day to day jobs, making them as easy as possible. In this paper, we present a review of the usage of these container technologies in the edge computing systems. In domains with constant data flow with processing on the very edge of the network.

1. On container usability in large-
scale edge distributed system
Miloš Simić, Milan Stojkov, Goran Sladić and Branko Milosavljević
Faculty of Technical Sciences, Novi Sad, Serbia

2. What is edge computing
• Nowadays we are facing a massive shift away from the standard,
centralized computing model that is provided through cloud computing
paradigm.
• The shift is now towards the distribution of computing power back to
the edge of the network.
• The basic concept of edge computing is to leverage new generation
technologies, processes, services, and applications that are built to
take an advantage of new infrastructure.

3. What is edge computing
• Separate geographic area where sensing
systems are, in 'micro data-centers' that collect
and (pre)process data
• Extending Cloud computing paradigm
• Try to overcome idea “one size fits all”
• Help Cloud in new use cases and applications
like autonomous cars, delivery drones, sensing
systems etc.

4. Edge computing challenges
• Service deployment
• Service management
• Service robustness and recovery plan
• Data caching
• Service monitoring
• Service configuration

5. Why containers
• Containers are not new technology
https://www.slideshare.net/insideHPC/linux-container-technology-101

6. Why containers
• Containers are usually compared to virtual machines
(VMs)
https://blog.netapp.com/blogs/containers-vs-vms/

7. Why containers
• Containers are based on few technologies that
allow various kind of isolations
• Containers allow users to run their services
isolated from the rest of the application virtually
anywhere and in a predictable manner
• Technologies that allow this behavior are:
- cgroups (resource restriction)
- namespaces (resources ‘dedicated' to the processes)
- unionfs (shared immutable layers, save space)

8. Edge applications
• Similar research show positive impact running applications in the
containers, we propose that multiple arm machines can be connected
and form micro datacenter
• This data center will mimic real datacenter but running specific
applications inside containers
• These applications are specifically designed for the edge computing, and
they will do filter and preprocess of the data before sending it to the cloud
• Edge applications come in form of 1) streams, that should continually do
some processing on data as it is arriving(long-running jobs), 2) standard
batch jobs that do some batch processing over some collection of data
and 3) reacting only on if some value passes some threshold in form of
events.

9. Pros and cons
• This model could be used to help cloud computing by
filtering and preprocessing data at the edge on the
network, and only send valuable data to the future analysis
• This accelerates time to market for the users and also save
money on storing unnecessary data
• But the problem with this approach is that we inherit all the
problems that could possibly come with the containers
• Not like real data centers, where we can choose between
VM, containers or some hybrid solution, and use best for
the specific use case

10. Conclusion
• We investigate does containers architecture
benefit edge computing especially in the large-
scale distributed environment
• We proposed possible application types that
could run in edge computing systems

11. Future work
• Research and offer a possibility that these edge
application could be run like unikernles
• Combine unikernes with container tools
• Lower the security risks that containers bring

12. Thank you for your
attention
‘Containers are great, let’s run them everywhere’
Brian Dorsey, Google