[Edge-computing] Classifying edge-site scenarios / Identifying expected features

Gnep Zhao jass.zhao at gmail.com
Fri Jan 12 02:24:42 UTC 2018

The infra layer (compute, not storage and network) has two type of runtime
technologies: container and virtualization. To provide a universal and portable
platform for applications, the underlying runtime is an implementation choice.
The package format, aka Image, is the key. Application should be packaged only
once in a "standard" format, and be deployed on any runtime, based on various
With that said, not all apps can be "dockerized" or turned into "micro-service",
some are built with the notion of Server. But that doesn't necessarily mean that
it has to run as VMs. LXD is able to provide a VM-like experience on top of
cgroup/namespace; and Kata is able to provide a Docker experience on top of KVM.
The myth is that Docker image has to be microservices. In fact, we can build
"fat", full server-like Docker images. But uniting both container and
virtualization under one single, de-facto image format and workflow has
tremendous benefits.  

On Fri, Jan 12, 2018 2:41 AM, Pasi Vaananen pvaanane at redhat.com  wrote:
Cannot comment on "some", but it is generally requirement for "many". Containers
do not mean that the do not need multi-tenancy, but if you want to include "all"
cases vs. only "some", then it is a requirement. 


On 01/10/2018 09:55 PM, Gnep Zhao wrote:
Just wondering whether multit-tenancy is a hard requirement for some Edge
environment, e.g. device, gateway, NFV appliance? Those setup sound more like
single tenancy, therefore container + baremetal is good enough?  

On Wed, Jan 10, 2018 10:17 PM, Allison Randal allison at lohutok.net  wrote:
The typical pattern for unikernels is to deploy the application

workloads as unikernels rather than as containers or VM/cloud images. (I

also pondered the possibility of trying the OpenStack services deployed

as unikernels, as I read the article, but that was more idle curiosity,

since it wouldn't buy us much in an edge scenario.)

I also wasn't suggesting directly trying the LightVM implementation in

OpenStack, since it's mostly a proof-of-concept at the moment. (Though

I'll be interested to see where they go with it.)

Mainly, I was suggesting that their workflow for performance analysis

and optimization of Xen is a good example of what we could do with

OpenStack Edge deployments, once we've defined a small set of

representative edge-site scenarios. That performance analysis workflow

applies equally well to either the small-footprint or distributed

approach to using OpenStack for edge infrastructure, you just have to

use a more distributed set of performance analysis tools for the

distributed case.


On 01/09/2018 10:38 PM, Paul-Andre Raymond wrote:

> Allison, 


> Thanks for sharing. I was not aware of this work and I found the article

> Today we discussed two aspects:

> 1- How to make openstack fit into smaller footprint (i.e. into an edge site)

> 2- How to make openstack more distributed (with Compute nodes at edge sites
connected via a WAN link)


> This article certainly fits in the first category. 

> In particular, if we think about deploying Openstack components in Unikernels
instead of deploying them with Containers.



> Paul-André

> --



> On 1/9/18, 6:00 PM, "Allison Randal" <allison at lohutok.net>  wrote:


> Following up on the conversation in the call today, have folks on the

> list read "My VM is Lighter (and Safer) than your Container"[0]? It's a

> good example of the kind of work we could do now on OpenStack, if we

> select a small number of representative use cases/scenarios, build

> sample deployments, and share concrete data with developers on specific

> changes we need in OpenStack. This kind of work can be ongoing at the

> same time as higher-level and broader-scope conversations about areas

> where we're still unsure of the best architecture or implementation details.


> Allison


> [0] Filipe Manco, Costin Lupu, Florian Schmidt, Jose Mendes, Simon

> Kuenzer, Sumit Sati, Kenichi Yasukata, Costin Raiciu, and Felipe Huici

> (2017) "My VM is Lighter (and Safer) than your Container." In

> Proceedings of the 26th Symposium on Operating Systems Principles (SOSP

> '17). ACM, New York, NY, USA, 218-233. DOI:

> https://doi.org/10.1145/3132747.3132763, Also available at:

> http://cnp.neclab.eu/projects/lightvm/lightvm.pdf


> On 01/05/2018 06:53 PM, lebre.adrien at free.fr  wrote:

> > Hi, 

> > 

> > Following the LTE discussion, I'm wondering whether we should not try to
classify the edge-site scenarios by level of the complexity (i.e.,
features/capabilities each scenario implies/requires).

> > 

> > A possible classification can start with: 

> > a) An edge infrastructure composed of several edge sites operated by a same
organisation with WAN wired connections. 

> > b) The same infrastructure but with LTE connections

> > c) Scenario a) but where the edge infrastructure is spread over several edge
sites belonging to different organisations/operators

> > d) scenario c) but with LTE

> > e) ..

> > 

> > In parallel, we can try to identify w.r.t these scenarios what are the
expected features/services from the administrator viewpoint and then from the
developers viewpoint.

> > 

> > I have the feeling that we all have relevant scenarios with specifics
according to our targets. Having a classification can allow us to move forward
by just focusing on classes instead of each particular scenario independently. 

> > 

> > My two cents, 

> > ad_ri3n_

> > 

> > _______________________________________________

> > Edge-computing mailing list

> > Edge-computing at lists.openstack.org

> > http://lists.openstack.org/cgi-bin/mailman/listinfo/edge-computing

> > 


> _______________________________________________

> Edge-computing mailing list

> Edge-computing at lists.openstack.org

> http://lists.openstack.org/cgi-bin/mailman/listinfo/edge-computing




Edge-computing mailing list

Edge-computing at lists.openstack.org


============================================================  Our greatest
contribution to the world is our dissatisfaction of everything  

Edge-computing mailing list
Edge-computing at lists.openstack.org

============================================================Our greatest
contribution to the world is our dissatisfaction of everything
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/edge-computing/attachments/20180112/9922fd3b/attachment.html>

More information about the Edge-computing mailing list