Hi Marcello,

In my experience, your suggested specification should be fine for use in a lab 
trail, in which you just place occasional calls. However, if you plan to send 
even low levels of load through the system, I'd recommend using the larger spec 
with 4GB RAM per node and 2vCPUs. This would require 28vCPUs and 56GB RAM - 
nearly all of the resources on your planned host. I recommend against 
over-provisioning resources for virtual telephony applications, as this can 
lead to unpredictable processing delays, and high latency.

One more suggestion. You are planning on using redundant Project Clearwater 
nodes in your deployment for resilience. However, you are still vulnerable to 
hardware failures as your host is a single point of failure. For a fully 
redundant system, you need at least two hosts, and you will need to set up 
anti-affinity rules to make sure that each of your pairs of nodes is split 
between the two hosts.

Given both of these points, I suggest adding a second host to your deployment.

Yours,

Chris

From: Marchitto Marcellino [mailto:[email protected]]
Sent: 24 March 2016 13:13
To: Chris Elford (projectclearwater.org) <[email protected]>; 
[email protected]
Subject: R: [Project Clearwater] HW requirements for clearwater vIMS nodes 
deployed on OpeStack Cloud

Hi Chris,

firstly I want to thank you for quick reply and professionalism. Thank you for 
the answers.

So I'd want to add a bit about my intents and get more suggestions from you 
about Clearwater.

I'm an IMS Engineer and, unfortunately, I discovered Clearwater too late :). My 
idea is to create an environment to test an opensource orchestrator for basic 
ETSI use cases, scale up/down in/out the Clearwater vIMS VNFs, simple calls and 
basic IMS services. I'm thinking to implement software OPNFV architecture 
(Brahmaputra release) for this purpose and my opinion to dimension this 
TestPlant is to have the following:

Resources:

·         2 x 8 core CPU > 2.00 Ghz (Intel VT-X o AMD-V Capable)

·         64 GB RAM

·         2 x 500GB SAS HD (Raid 1/0)

NFVI will be created with OpenStack as Iaas infrastructure, KVM as hypervisor 
and Compute node, ONOS, OVS, OpenDaylight and OpenContrail for Network node, 
Ceph for Storage node. Then DPDK and ODP as Accelerators.

So the following architecture:
[cid:[email protected]]

Moreover a MANO open software like cloudify (for example) to manage Onboarding, 
VNFs lifecycle, scale up/down/in/out, multiple VNFMs and VIMs ecc...

About number of VMs I think:

Virtual Machines:

·         7x2 VMs (Sprout, Bono, Homer, Ellis, Bind9, Homestead, Ralf) in a 
redundant configuration;

·         1 VMs for Openstck;

·         1 VM for Ceph;

·         1 VM for ONOS, OVS, OpenDaylight and OpenContrail

·         1 VM for OpenMANO or Cloudify

For the moment I'm thinking to get experience only on Openstack cloud, then 
I'll move to VMWare too.


·         Do you think these Resources and Virtual Machines are enough for my 
scope? What do you think ?

·         Do you see some limitation by Clearwater vIMS node side?

I'm sorry if I ask you so many things but your product is very interesting as 
testing purpose on NFV. I'm very grateful if you could answer to these my 
doubts.

Thank you very much,

I'm looking forward to hearing you.
Best Regards,
/Marcello

Da: Chris Elford (projectclearwater.org) [mailto:[email protected]]
Inviato: giovedì 24 marzo 2016 10:48
A: Marchitto Marcellino; 
[email protected]<mailto:[email protected]>
Oggetto: RE: [Project Clearwater] HW requirements for clearwater vIMS nodes 
deployed on OpeStack Cloud

Hi Marcello,

Thanks for getting in touch.

As part of Project Clearwater development, we have tested on OpenStack, VMware, 
and AWS. We have used KVM for our OpenStack testing as it's the default, but 
other hypervisors should work.

We usually use small VMs. Our chef-created deployments in Amazon EC2 use VMs 
with 1vCPU, 2GB RAM, and 5GB storage. We have also done some testing on VMware 
using larger VMs: 2vCPUs, 4GB RAM, and 20GB storage. Project Clearwater scales 
by adding more nodes to the cluster, not increasing the size of the individual 
VMs, so these should be a reasonable place to start.

Please let us know if you have any more questions.

Yours,

Chris

From: Clearwater [mailto:[email protected]] On 
Behalf Of Marchitto Marcellino
Sent: 22 March 2016 23:58
To: 
[email protected]<mailto:[email protected]>
Subject: [Project Clearwater] HW requirements for clearwater vIMS nodes 
deployed on OpeStack Cloud

Goodmorning clearwater project,

I'm contacting you to have more information about Hardware Requirements of your 
vIMS Architecture. In order to deploy it in our Lab based on OpenStack Cloud 
manager I'd like to get an answer for the following items:


*         RAM, CPU, Storage for any Virtual Machines related to IMS nodes on 
large-scale deployment;

*         Hypervisor type

On website I got something like this: 2GB, 1vCPU (1GHz since Ubuntu), more then 
1,75GB HD(since Ubuntu).
No info about Hypervisor compatibility.

Cloud you help me please?

Thank you very much,
Best Regards,
/Marcello

Marcellino Marchitto
Telecommunication & Media
BU Telco&Media
Adv. Consultant
Via Tiburtina, 1232
00131 Roma
Italy
Mob. : +393426959875
[email protected]<mailto:[email protected]>
www.altran.com<http://www.altran.com/>

Please consider the environment before printing.
"This message and any attachments (the "message") are confidential and intended 
for the sole use the address(es). Any unauthorised copy or dissemination is 
strictly prohibited. If you are not the intended recipient, please notify us 
promptly and delete such message from your inbox."


_______________________________________________
Clearwater mailing list
[email protected]
http://lists.projectclearwater.org/mailman/listinfo/clearwater_lists.projectclearwater.org

Reply via email to