how it routes and network question

Hi

I recently get two dedicated servers in hosting company

the ip address of the server is eg: 192.168.32.4/32 and 192.168.57.2/32

I try to check the gateway but it seems no any trace command in the server.
hosting company is using dhcp in those servers network configuration.

I have concern and questions

1/ How can I know those servers are same network as those net mask is /32
if not the same network, I have problem when two servers are slowing
in transferring files

2/ lf the network card in server has problem and need change another
one, will my ip address change to another ip address also?

3/ why hosting company is using /32 and dhcp? what is advantage? ls it
easy for administration?

Thank you for your help

<SNIP>

1/ How can I know those servers are same network as those net mask is /32
if not the same network, I have problem when two servers are slowing
in transferring files
</SNIP>
Check the subnet mask, i doubt its going to be 255.255.255.255

<SNIP>
2/ lf the network card in server has problem and need change another
one, will my ip address change to another ip address also?
</SNIP>
Yeah well thats how dhcp works, via ma caddy, i guess you can always spoof
your old mac address.

<SNIP>
3/ why hosting company is using /32 and dhcp? what is advantage? ls it
easy for administration?
</SNIP>
Im guessing because the users are to stupid to understand what a subnet
mask/gateway is its just easier to get the mac address and assign it to a
user then let the user assign a ip.

Normally in a co-location setup its not like this, inless its very cheap
hosting.

My co-location has the following setup, and this is how MOST networks should
be run.

Core router using BGP to transit providers, and other local peers.
Switched network useing ospf to handle the routes and also VLAN's for the
customers subnets.

So customer should get a vlan assigned to them (which they have no need to
know what the number is, they are handed a access mode port.
Customers also issued a /30 (at least) in most cases a customer will get a
/29 or /28 depending on what they need.
In this case of a /30 its a total of 3 address's
1, GATEWAY (put on the ISP/HOST switch
2, IP ADDRESS FOR SERVER TO USE
3, BROADCAST ADDRESS.

Heres an eg of a /30:

Address: 192.168.1.1 11000000.10101000.00000001.000000 01
Netmask: 255.255.255.252 = 30 11111111.11111111.11111111.111111 00
Wildcard: 0.0.0.3 00000000.00000000.00000000.000000 11
=>
Network: 192.168.1.0/30 11000000.10101000.00000001.000000 00
HostMin: 192.168.1.1 11000000.10101000.00000001.000000 01
HostMax: 192.168.1.2 11000000.10101000.00000001.000000 10
Broadcast: 192.168.1.3 11000000.10101000.00000001.000000 11
Hosts/Net: 2 Class C, Private Internet

Heres an eg of a /29:

the % ipcalc 192.168.1.1/29
Address: 192.168.1.1 11000000.10101000.00000001.00000 001
Netmask: 255.255.255.248 = 29 11111111.11111111.11111111.11111 000
Wildcard: 0.0.0.7 00000000.00000000.00000000.00000 111
=>
Network: 192.168.1.0/29 11000000.10101000.00000001.00000 000
HostMin: 192.168.1.1 11000000.10101000.00000001.00000 001
HostMax: 192.168.1.6 11000000.10101000.00000001.00000 110
Broadcast: 192.168.1.7 11000000.10101000.00000001.00000 111
Hosts/Net: 6 Class C, Private Internet

Hope this makes sence.

Regards,

Bruce

Hi Bruce

Thank you so much to explain me in detail. I would like to know about
this it in case i can get another hosting company

Yes. I think the netmask should be 255.255.255.255
1/ but why they are using this netmask setting? save ip address?
then does the router handle many routes in this setting?
2/ What is this advantage for the hosting company?
3/ If I need more ip in the same server, how it works?
4/ Why you said the hosting company is cheap to use this configuration?

Thank you again.

Hi, your "hosting company" is likely NAT'ing or using load balancers on the front end. You are obviously not "reaching" those machines by ssh'ing into 192.168.x.x... Additionally, assuming that DHCP is handing out that address on the server that mask would likely not be all ones.

Even Amazon EC2 instances use private addresses now on the backend ...

Kind regards,
Truman

<SNIP>
Yes. I think the netmask should be 255.255.255.255
1/ but why they are using this netmask setting? save ip address?
then does the router handle many routes in this setting?
</SNIP>

I have no idea the only way you can have a /32 is with a ppp that doesn’t
use arps to talk to each end of the tunnel.
I would assume they have /24's and are giving out /32 via dhcp to customers
and the customers should see 255.255.255.0 with a gateway of eg, .1 etc..

<SNIP>
2/ What is this advantage for the hosting company?
</SNIP>

If the company is setup the way i think it is the only reason for this is:
1, they have no clue what they are doing
2, they offer a very cheap hosting service and have no managed switches, and
don’t understand how to subnet and use vlan's.

<SNIP>
3/ If I need more ip in the same server, how it works?
</SNIP>

I would of thought if you have 2 x servers you wanted to co-locate the
hosting company would offer you a /29 with 1 gateway 1 broadcast and 4
useable on a vlan, so local traffic only sits on the vlan and the servers
can talk to each other via the local vlan.

I guess if the machines have more then 1 nic you can connect the 2 machines
via a local 'backnet' network it can be useful if you have a cross-over
cable between the 2 x machines and its a 1GB port. This also saves using the
switches, in some cases hosting companies may count all traffic that goes
over the interface (if they don’t use net flow) and you could end up paying
for traffic which you really shouldn’t have to pay for.

If you are using the additional ports for high amounts of data eg, backup's
images etc, you can really tweak tcp settings so you can send JUMBO frames
and squeeze some speed out of it.

<SNIP>
4/ Why you said the hosting company is cheap to use this configuration?
</SNIP>

Yes its alot cheaper to have say a common-gateway that all traffic will
route over and then connect a bunch of switches to this common router and
manage it via dhcp, its very messy and also very noisy i can only imagine
after you connect a few servers that over time you will see arp storms and
all traffic on the network will cease to flow.

As mentioned in my other posts how it should be done, clearly you need to
buy layer 3 switches and layer 2 switches and a nice core router to deal
with your bgp, you also need to make sure your using devices that can handle
high packets per second.

As i am writing this i feel as if im doing someone homework for them... :stuck_out_tongue_winking_eye:

Thank you again.

<SNIP>
2/ lf the network card in server has problem and need change another
one, will my ip address change to another ip address also?
</SNIP>
Yeah well thats how dhcp works, via ma caddy, i guess you can always spoof
your old mac address.

<SNIP>
3/ why hosting company is using /32 and dhcp? what is advantage? ls it
easy for administration?
</SNIP>
Im guessing because the users are to stupid to understand what a subnet
mask/gateway is its just easier to get the mac address and assign it to a
user then let the user assign a ip.

Normally in a co-location setup its not like this, inless its very cheap
hosting.

My co-location has the following setup, and this is how MOST networks

should

I should add; i guess i made some assumption that you were co-locating your
own servers with someone, if this isn't the case, please ignore everything
i'v said :wink:

-bruce

Or these are VPS', and not physical Servers.

From my brief encounters with various VPS technologies, this makes more

sense.

Regards,
Kieran.