<div dir="ltr">Hi Hossein,<div><br></div><div>This is a pretty open-ended question that can be answered several different ways.</div><div><br></div><div>Have you read through both the <a href="http://docs.openstack.org/openstack-ops/content/">Operators Guide</a> and <a href="http://docs.openstack.org/arch-design/content/">Architecture Design Guide</a>? Both of those docs might contain some useful advise. There's also a <a href="http://docs.openstack.org/high-availability-guide/content/index.html">High Availability Guide</a>, though I believe it's going through a rewrite at the moment.</div><div><br></div><div>My personal opinion for three nodes is to allocate one node for the controller services (keystone, glance, nova-api, etc) and the other two for compute. When you're able to add more nodes to the cloud, dedicate one as the second controller to make things HA.<br></div><div><br></div><div>You should be able to add more controller services with minimum service interruption. I'm not too familiar with Neutron, but from personal experience, all other OpenStack components scale horizontally very well behind a load balancer.</div><div><br></div><div>If you want to prep your single controller for expansion, begin by installing a load balancer such as HAProxy on the single controller to pass all requests to the services running on that same controller. Then when you get a second controller, you can simply append the new services to the HAProxy config. You can then introduce something like VRRP or uCARP to balance a shared IP between the two controllers.</div><div><br></div><div>Hope that helps,</div><div>Joe</div><div><br></div><div><br></div><div><br></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Thu, Oct 23, 2014 at 4:24 AM, Hossein Zabolzadeh <span dir="ltr"><<a href="mailto:zabolzadeh@gmail.com" target="_blank">zabolzadeh@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Appreciated if someone help to find my answer.<br>
Thanks in advance.<br>
<br>
On 10/22/14, Hossein Zabolzadeh <<a href="mailto:zabolzadeh@gmail.com">zabolzadeh@gmail.com</a>> wrote:<br>
> Any idea?<br>
><br>
> On 10/21/14, Hossein Zabolzadeh <<a href="mailto:zabolzadeh@gmail.com">zabolzadeh@gmail.com</a>> wrote:<br>
>> Hi there,<br>
>> I have an openstack deployment question. I have three physical<br>
>> server(Configurations are listed in the following). What is the best<br>
>> optimized production-ready deployment model to achieve the following<br>
>> goals in mind(Users will not exceeds more than 10)?<br>
>> - Ability to add more services(compute, controller, storage node)<br>
>> later, without service interruption<br>
>> - Ability to provide HA for all services, after more nodes added<br>
>> without service interruption<br>
>> - No need to change the deployment architecture later when more<br>
>> physical nodes and resources available(My physical infrastructure will<br>
>> be scaled later, and not limitted to these three physical machines).<br>
>> For example I don't know is it better to have 1 neutron node, or have<br>
>> two neutron service on my two compute nodes?<br>
>> A really appreciate any helps.<br>
>> Thanks in advance.<br>
>> ------<br>
>> Server 1:<br>
>> RAM: 128GB<br>
>> HDD: 300GB<br>
>> ------<br>
>> Server 2:<br>
>> RAM: 128GB<br>
>> HDD: 300GB<br>
>> ------<br>
>> Server 3:<br>
>> RAM: 256GB<br>
>> HDD: 512GB<br>
>><br>
><br>
<br>
_______________________________________________<br>
OpenStack-operators mailing list<br>
<a href="mailto:OpenStack-operators@lists.openstack.org">OpenStack-operators@lists.openstack.org</a><br>
<a href="http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators" target="_blank">http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators</a><br>
</blockquote></div><br></div>