Hi guys
I started to work in a small company which doesn't have infrastructure and I need to built it.
My problem is: I am a student with theoretical knowledge but without practical one so I don't have any ideas how to solve my problem.
I installed on CentOS 6.4 DNS and DHCP. This two worked well untill my boss told me to create two subnets. On the first subnet (192.168.1.0/32) we have out workstations and this is usually network on which our partners can connect. On subnet 10.0.0.0/32 I have servers which we need for our work.
Problem is: I don't know to set up corporate network so I can ping servers on subnet 10.0.0.0/32 from subnet 192.168.1.0/32. We don't have any fancy routers and all my servers are virtual, running inside Virtualbox.
On the same CentOS, I installed and configured DHCP and reserved addresses for all my servers, enabled routing, turned off firewall for now and add manually route on my Windows workstation. I can ping server on other subnet by IP address but not with hostname.
Please, help me.
DNS is about 3 things, finding Domain servers, reverse lookup of all IPs including local, and forward lookup of hosts including local. The last part is divided into twoareas, dynamic supporting DHCP and fixed for servers. The forward and reverse domains need to be known to the parent domain. Reverse IP lookup is done using domain inaddr.arpa.IP, so in addition to servers being installed in the local domain, somehow all IPs have to be installed into inaddr.arpa.
New subnets is about configuring them on a router so they get routed to. Manual routes is a hack. Giving them reverse DNS support is also nice. But do not otherwise confuse domains and subnets. One is in the Name space, the other in IP address space. DNS lookup is hierarchical through name servers, although local overrides are possible (I once configured local DNS so aol.com was our firewall, where there was a proxy to the real AOL.) and the IP routing world, handled by routers. Now, any host with two NICs is a potential (when IP forwarding is turned on) router. Best practices is to use dedicated routers, so non-router hosts just point to the router(s) at the head of their subnet. Routing protocols or manual settings on routers establish the routes for all hosts to and from your hosts. You could use any box as a router (even if it is also a server):
- configure it with 2 or more NICs, for each child subnet and the parent subnet,
- IP Forwarding turned on and
- higher level router(s) as its default route
- set the child subnet hosts to see it as the default route,
- To enable routing in, either:
[list]
- install and activate a routng protocol compatible to the parent router(s) (you can look for their packets to discover what is running) or
- set the parent routers manually to see it as the route for your subnets.
[/list]
When hosts send it traffic, it will forward to the a local subnet or up to its default.
For redundancy, my guess is that you do it twice. I suspect a host with two routes will use one until it stops working and then try the other. But there are many ways to do this, like two hosts using the same IP but only if the other is not, like a cluster.