Services that need a static IP

ElkyElky

I'm working on a project and hoping someone could help me with something.

I'm compiling a list of services (software or hardware) that requires the use of a static IP in order to function properly. Anything that can't function correctly using NAT and needs its own dedicated IP. Can anyone name some?

I know that a website with SSL needs a static IP. Perhaps a hardware firewall or VPN?? Haven't really found much on Google so hoping the geeks here could help me.

Thank you!

jbainbridge

- web server
- email server

Strider590

I don't have a static IP (in fact I have my modem on a timer socket to deliberately rotate my IP every day), BUT I run a web server and an FTP server, both of which I access on a daily basis.

I simply use a dynamic DNS provider.

esuhl

If you disable uPnP on your network, you may need to assign static IPs to any devices that need port-forwarding.

John_Gray

On your own local LAN it is often a good idea to assign a static IP to a networked printer, rather than allowing DHCP to assign its IP address.

OGR

Domain controllers should always have a Static IP in a Windows domain.

Management controllers on Servers should be assigned static IPs so that you can find them easily such as the iDRAC on a Dell server.

Managed Switches should be assigned a static IP.

Any virtual servers and management systems (such as vSphere vCentre).

Each NIC on a SAN/NAS.

There is tons of stuff in the business environment that you would want to assign a static IP, usually anything you run on or as a server.

Big_Graeme

Are you talking about private IP addresses of public ones?

For instance a web server will work fine on a variable ip as long as you have a DDNS service set up.

ElkyElky

Helpful replies, thanks!

I'm writing about IPv4 exhaustion... and if I were looking at from an ISP/Ripe perspective. Obviously there are strict rules for assigning services/hardware a static IP given the non-existent supply, and Ripe would only grant someone a block of addresses for specific needs where it would only be essential.

So if I were working for Ripe, what things would I consider a valid reason for requiring its own dedicated address?

This is trickier than I thought it would be!

onomatopoeia99

By 'static IP' you mean a unique, unchanging, routeable IPv4 address then (no RFC1918 allowed)?

I would say HTTPS on the standard port, as you can only have one certificate per IP address and port combination, so no name based virtual hosting (which is possible on HTTP). However, there is nothing to stop the machine running the HTTPS service being on an internal network with the port forwarded through a NATing router to a real IPv4 address.

So, are you asking about services that you can only have one of per real IPv4 address or services that actually break if NAT is involved?

FTP breaks, as it sends internet layer data (the IP address) in application layer packets. This means the device doing NAT, an internet layer action, has to be aware of the structure of application layer data.

SIP (used by VOIP) breaks in similar ways, and if it's using encryption so the NAT device can't see the address information in the application layer data then it can't be corrected.

If I develop a new application layer protocol or a new transport layer protocol, it will break on NAT as no extant devices will be aware of it. This means NAT fundamentally hampers protocol development.

NAT is evil. Amazingly, people are talking about it for IPv6. Really? It has enough address space that every grain of sand could have its own unique address, it does not need private address spaces.

Since I stand by my "NAT is evil" statement with my actions, all of my devices are on proper, routeable, non-RFC1918 IPv4 addresses. Even my TV and my Blu-Ray player (and my smart phone when connected to my wifi).

I would suggest that people who want to run internet servers and therefore need unique IP addresses for them should use IPv6 nowadays. No excuses, it's been around for ages.

If I was RIPE I'd just allocate them all on request until they were gone and tell everyone else "IPv6 only now." This probably doesn't help you much though!

I_have_spoken

IP4 running out of addresses and static IP are not the related

securityguy

ElkyElky wrote: »

Helpful replies, thanks!

I'm writing about IPv4 exhaustion... and if I were looking at from an ISP/Ripe perspective. Obviously there are strict rules for assigning services/hardware a static IP given the non-existent supply, and Ripe would only grant someone a block of addresses for specific needs where it would only be essential.

So if I were working for Ripe, what things would I consider a valid reason for requiring its own dedicated address?

This is trickier than I thought it would be!

The UK ISPs have plenty of IP numbers, to the point that the boutique ISPs will give even residential customers a /29 or even a /28 if you ask them nicely. Your mention of RIPE makes me think you're confusing IP numbers an ISP would allocate within their space with directly allocated provider-independent IP space, the sort of thing you would link to an autonomous system number and advertise with BGP, which means a minimum of a /24, which no ISP can give you.

In the early 1990s my ISP gave me a /21 and a /22, a total of over 3000 IP numbers, for a 500 employee company, and I already had four /24s from one of RIPE's precursors, for a further 1000. And in the mid 1980s I acquired a /16 for a small computer science department with no need to do more than ask: 65536 IP numbers when we probably had about fifty computers in total, and at the same time the central computer service acquired another /16 and I think physics likewise. Even today, a large Russell Group university probably has at least 1 IP number per student and staff.

But now, getting a /24 of provider-independent space from RIPE would require a very serious justification, probably about multi-homing.

It's easier to just use IPv6. I've got a /48 at home, so unless it turns out I need 1208925819614629174706176 computers, I'm OK...