Category Archives: Networking

Networking

In-Wall Cable Ratings

When installing some coaxial (RG6) cable in a home, you really don’t have too many things to worry about. No fire marshal is going to fine you for having the wrong type of cable installed.

At a business, on the other hand, you have quite a few things to worry about:
Is the cable going to be installed in the wall or just be out in the open?
Is it going to be installed in a plenum or a standard ceiling riser?
http://www.bluejeanscable.com/articles/inwallrating.htm – THE place to check all this information out.

I won’t bother copying everything, but here’s the gist of it:
CMP – Rated for all in-wall, in-riser, or in-plenum applications
CMR – Rated for all in-wall, in-riser applications
CM – Rated for all in-wall applications

Why the different ratings? Plenum covered wires don’t give off a toxic smoke when burned. I assume that the in-riser cables take a bit more heat to give off the fumes and that standard CM rated cables don’t really help much against the spread of fire. In all of my residential and most of my commercial jobs I’ve installed CMR cable, and anytime I’ve installed in the plenum I’ve used CMP wire.

Networking 101

I’m going to try to make this as easy to understand as possible so bear with me. It’s a very geeky topic and I’m not usually a great writer, so I’ll try to not make it incredibly boring. Or long. Well, it’ll probably be long, but it should be a decent read.

OSI Model
First we have to at least look at the OSI (Open System Interconnection) model. You don’t have to completely understand it, but a few reference points would be mighty handy.
There are 7 layers to the OSI model:
1.) Physical Layer – The actual physical connection (modem cable, serial cable, ethernet cable, etc)
2.) Data Link Layer – Provides the functionality and means to transfer data (MAC Addresses)
3.) Network Layer – IP layer (TCP and UDP)
4.) Transport Layer – Provides reliability controls to the layers above (QoS)
5.) Session Layer – Mechanism for opening and closing sessions
6.) Presentation layer – Encryption layer, also responsible for formatting data for the Application layer
7.) Application Layer – Final layer in which applications (Web browser, Email client) take the formatted data and display it for the user

Keywords
Next I’d like to try out keywords – just to see if you know what is what.
ICMP – Internet Control Message Protocol, or Ping, is used to send error messages regarding host connectivity
Packet – Payload of formatted data sent by network devices
MAC Address – Media Access Control address is a unique physical ID embedded on every network device. It’s a 48bit address of six groups of 2 digits separated by hyphens or colons (01-23-45-67-89-ab)
IP Address – Internet Protocol address is a numbered 32bit address consisting of four groups of 3 digits separated by periods (127.0.0.1 and 192.168.1.1)
Switch – a switch is a device with multiple ethernet ports that connects network devices
Router – Chooses paths for packets and connects different networks together
Ethernet – Patch cable, consists of 8 wires with an RJ45 connector. It’s like a phone cable on drugs.
QoS – Quality of Service. Allows for certain traffic to be flagged at higher priorities than other traffic. Usually Servers beat out clients, and IP Phone conversations beat out web browsing.
Duplex – Full or half duplex. Full duplex means a network connection can send AND receive at the same time at the same speed. Half duplex is a walkie-talkie system in which the system must wait for a free moment to speak.
WAN/LAN – Wide area network/Local area network. WAN refers to an internet/cloud connection, LAN refers to any locally connected devices. WLAN refers to Wireless Local Area Networks.
NAT – Network address translation. Allows for several systems to use the same public IP address at the same time.
Ports – a connection on a switch, hub, or router.
TCP Ports – Ports are added onto TCP/UDP packet headers for destination use. Ports range from 0 to 65535. Port 80 is used for web browsing.

I already know that I will have to write up another one for IPv6 information. This is only for IPv4. IPv4 addresses are said to be exhausted by June 2011.

What is an IP address?
The basic building block of the internet is the IP address. IP addresses (IPv4 anyway) are 4 sets of numbers 0 through 255 separated by a ‘.’ Examples include 4.2.2.2, 192.168.1.1, 172.16.40.254.
There are Class A, Class B, Class C, Class D, and Class E networks (for all of our purposes you only have to remember AB and C).
Class A: 128 available networks each having 16,777,216 IP addresses. Start address is 0.0.0.0 end address is 127.255.255.255
Class B: 16,384 available networks each having 65,536 IP addresses. Start address is 128.0.0.0 end address is 191.255.255.255
Class C: 2,097,152 available networks each having 256 IP addresses. Start address is 192.0.0.0 end address is 223.255.255.255
Class D: Used for multicast on networks, reserved for private use. Start address is 224.0.0.0 end address is 239.255.255.255
Class E: Reserved but not in use. Start address is 240.0.0.0 and end address is 255.255.255.255
Unfortunately not all IP address can be used – there are quite a few that are reserved for private purposes:
0.0.0.0 through 0.255.255.255 is known as the Zero Addresses and cannot be used currently.
10.0.0.0 through 10.255.255.255 is a private IP address scheme for Class A. These are unroutable IP addresses (they only exist in private networks).
127.0.0.0 through 127.255.255.255 is a localhost/loopback address range. 127.0.0.1 is always set as the localhost and can not be routed to the internet.
169.254.0.0 through 169.254.255.255 is for zeroconfig addresses when a DHCP server is not working or present on the network.
172.16.0.0 through 172.31.255.255 is a private IP address scheme for Class B. These are unroutable IP addresses (they only exist in private networks).
192.168.0.0 through 192.168.255.255 is a private IP address scheme for Class C. These are unroutable IP addresses (they only exist in private networks).
224.0.0.0 through 239.255.255.255 is reserved for multicast use. These are unroutable IP addresses.
240.0.0.0 through 255.255.255.255 is reserved and not used in the routing of internet IPs.

What the heck is Netmasking?
Netmasking is a way to differentiate between networks. This is also called subnetting. In the early days of the internet, if a company needed more IP addresses than a Class B network (say 67 thousand IP addresses were “needed”), they’d have to get either 2 class B networks (need 67 thousand, given 131 thousand) or a class A network (need 67 thousand, given 16 million). Now we have a way to lower the number of IPs given out for each class.
You may have seen the netmask 255.255.255.0 before – it’s the most common for home networks. You probably have an IP address of 192.168.1.50 with a netmask of 255.255.255.0. The 255’s mean that the network you are currently on must match. So your network is 192.168.1.0 through 192.168.1.255. If you had a subnet of 255.255.0.0 your network would be 192.168.0.0 through 192.168.255.255. If you had a subnet of 255.0.0.0 your network would be 192.0.0.0 through 192.255.255.255. See how that works?
And yes, you can divide it further – since there are only 2 million available class C networks we can divide them further:
192.168.1.0 with a netmask of 255.255.255.128 would mean that you have the network 192.168.1.0 through 192.168.1.127. If you have 192.168.1.128 with a netmask of 255.255.255.128 that would mean you have 192.168.1.128 through 192.168.1.255. That gives you 128 hosts.
Ah, but we can do better than that. What if you only want 16 hosts? 192.168.1.0 netmask 255.255.255.240. IP range would be 192.168.1.0 through 192.168.1.15. It’d also be a range of 192.168.1.16 through 192.168.1.31. Also 192.168.1.32 through 192.168.1.47. See how that works?
Each of these networks is different. In the last example with the netmask of 255.255.255.240, we see that all of the IP addresses have a 192.168.1.X start. But a computer with an IP of 192.168.1.4/255.255.255.240 could NOT ping a computer with an IP of 192.168.1.18/255.255.255.240 without the use of a router.

What’s so special about routers?
A router, or layer 3 switch, is required to connect different networks. If you were to run a trace route to apple.com, you will see that many routers are necessary for your packet to succeed. Here’s mine:
tracert
As you can see, my ping goes through the following:
A computer at 192.168.168.151 - My computer
A router at 192.168.168.1 - Local
A router at 68.85.1x.x - neighborhood
A router at 68.87.230.5x - Central Office
A router at 68.86.90.53 - Chicago backbone provider
A router at 4.71.248.17 - Chicago backbone provider
A router at 4.69.138.158 - Denver backbone provider
A router at 4.69.132.57 - San Jose backbone provider
A router at 4.69.132.157 - Sacramento backbone provider
A router at 4.69.132.149 - Sacramento backbone provider
A router at 64.158.148.6 - Apple Sacramento backbone provider
A router at 17.112.254.26 - Apple router
A router at 17.112.254.29 - Apple router
A webserver at 17.112.152.57 - Apple.com
That’s 12 routers just to get to apple.com. Impressive.
You’ll see that there are 12 different networks connected via a series of routers. None of this would be possible without these routers.

How does this all work?
MAC addresses, route tables, and a lot of configuration. Without getting too advanced (Keep It Simple Stupid KISS):
A switch will “remember” what port each MAC address is from. If the switch doesn’t know a MAC address source, it will write the MAC address and port number to it’s memory for future reference. If a switch doesn’t know the MAC destination, however, it will flood every port with the data. If no network responds, the switch simply throws out the packets of data.

A router will “remember” what port each MAC address is from. Remember that MAC addresses are physical addresses at the Layer 2. IP addresses are at Layer 3, so a conversion takes place:
Computers A and B are connected to a Switch. Computer A (192.168.1.2, MAC of 00-00-00-00-00-01) wants to send information to Computer B (192.168.1.3, MAC of 00-00-00-00-00-02). Computer A sends the packet to “192.168.1.3”. The switch looks in it’s table for that IP address. If it finds the IP, it then grabs the associated MAC address and finds what port that MAC is connected to. It then sends the data out that port. If the switch does NOT find the IP, it will send out the data to each port on the switch except the port it came from. If it gets a response, it will record the MAC address and port from the respondee. In our case, it’s computer B with 00-00-00-00-00-02.

What’s the difference between a router, a switch, and a hub?
A router routes, a switch switches, and a hub… well a hub just kinda sits there. Technically speaking, a Router will connect networks together, a switch will allow different network devices to communicate on the same network, and a hub extends the physical network. A hub is shared networking – if you have a 10/100 hub with 8 ports, each of the ports will share the total bandwidth of 100Mb. Hubs are half-duplex and are referred to as “dumb”. Switches are full-duplex and each port gets the entire bandwidth allotment. Many home routers also include a switch. A common example is a Linksys wireless router – it will include a WAN port, 5 LAN ports, and a pair of antennas.

So what about this layer 3 stuff?
The most common layer 3 protocols are TCP and UDP. While both protocols are basically the same in terms of the header/source/destination, they differ greatly in terms of usage. TCP is used when data integrity is required. Each packet of data is numbered and acknowledged. If a packet is “missing”, the computer will resend the same packet to the destination. UDP, on the other hand, has no integrity check. If a packet is received out of order it is discarded by the destination. Why would you want to discard out of order packets and not have the sender resend them? IP Phones, streaming video, streaming audio. Can you imagine talking on the phone and having parts of your conversation show up later? UDP is great for Live events – events where a delay is not accepted. TCP is used when integrity is required – grabbing email, downloading files, etc.

How about this “private” and “public” IP stuff?
Private IP addresses are those that can not be routed through the internet. Although technically you can use a “private range” of 17.112.152.x with a subnet of 255.255.255.0 on your home router, you would not be able to visit any apple websites. So, to keep things easier for everyone, the most common Private IPs are: 192.168.x.x, 172.16.x.x-172.31.x.x, and 10.x.x.x.

How is that different from your public IP? Most home users have a single Public IP, and usually it’s not statically assigned (meaning it can and will change if you ever lose your internet connection). Visit http://whatismyip.com sometime and see what your public IP address really is. We’ll assume that your private IP is 192.168.1.2 and your public IP is 17.112.152.57. A simple host lookup on the public IP will yield apple.com. If you were to ping apple.com, you will receive a response from the public IP (still 17.112.152.57). How can the two IP addresses be different? Because you have a router!

If you have a home network with 4 computers connected (2 laptops, a desktop, and an Xbox360), each has their own private IP (192.168.1.2, 192.168.1.3, 192.168.1.4, and 192.168.1.5 respectively), but they all share your public IP of 17.112.152.57. There is a process known as NAT that allows each of these devices to share a single public IP address. Every request sent outward to the internet gets modified by the router to have a unique source port. That way when the data comes back the unique source-turned-destination port will allow the router to “remember” which system requested the data. So if you’re browsing the web on your laptop while downloading updates on the xbox, there should be no conflicts of data.

Ports? What’s port 443?
Common ports are easy to find out. When you send for information from a website, it will run under port 80. If you want to request from a secure website, it will run under port 443. Common ones:
21 FTP
23 telnet
25 SMTP
53 DNS
80 HTTP
110 POP3
443 HTTPS
Your router will request the page on your behalf, sending the request to port 80 at apple.com. But it will inject port 57803 (random port number) as the source. When apple.com replies, the source is port 80 and their destination is your router at port 57803. The router then knows that you were requesting the information and forwards it onto you. Obviously this is done much more rapidly than I can type or you can read.

Hopefully that helps a little bit with the understanding of IPs and networking. I’ve only touched the surface when it comes to explaining all of this. And since I’m more of a visual learner, I’ll probably draw some pictures eventually.

Block DNS DD-WRT

Here’s the project:
Secure a wireless access point for vendors/non-work-computers to use. Disallow access to anything that eats up bandwidth – it’s supposed to be a tool and not a play thing.

Background:
A primary internet connection has 62 public IP addresses. One is pointed to a WRT54-G (v1) router. I installed dd-wrt (v24 preSP2 build 13064) and configured accordingly. I signed up for OpenDNS service to block all the “bad stuff”.

Problem:
Merely putting in the DNS information into the DHCP server is not enough. Any savvy tech user can add their own DNS information into the equation and go from there. I needed a way to block DNS (port 53) on the router side of things so that no outside influences could bypass the security. DD-WRT was configured to block port 53, but that didn’t really work out too well. So I ended up adding my own iptable chain to the router not to block other DNS, but to force it to go through the router’s DNS. Makes things easier in the long run.

Solution:

Log into the router
Click on the Administration Tab
Click on the Commands Tab
In the commands box, paste the following:
iptables -t nat -A PREROUTING -p udp -i br0 --dport 53 -j DNAT --to $(nvram get lan_ipaddr)
iptables -t nat -A PREROUTING -p tcp -i br0 --dport 53 -j DNAT --to $(nvram get lan_ipaddr)
Click Save Firewall

After it reboots, test it out. Use 4.2.2.2 as a test DNS server to verify. Awesome, I know.

Security:
WRT54-G running DD-WRT with a non-standard password
SSID is not being broadcast
WPA2 Personal with AES Encryption
WRT54-G is connected to a portion of the main internet pipe on a public IP but is VLAN’d off from the rest of the network
DD-WRT admin access is not allowed over the WLAN
OpenDNS is blocking almost all access
DD-WRT does NOT allow connections before 7AM, after 7PM, or on weekends
I’m wearing pants

OK, had to lighten the mood a little bit. But that’s the gist of it.

How To Reboot A Juniper Netscreen Firewall Via GUI

Sorry for the lack of posts lately – I recently got married, had a honeymoon, and was fired all within the last month. Stick around, I’ll be sure to post more when I get a new gig – until then I’ll just stick with my on the side consulting services.

I’m not too familiar with Juniper stuff as every place that I’ve been employed only used Cisco/less than Cisco gear. However, at a recent client I was given the task of rebooting a Netscreen that was acting up. Uptime on the netscreen50 was at 410 days (they don’t patch much), but I’ve heard of routers and switches going for several years between reboots. However, this router is an outward facing router with VPN support.

To reboot, log into the web interface (GUI stands for Graphical User Interface). Then navigate to Configuration, Update, ScreenOS/Keys. You’ll see at the bottom of this screen a “Reset” button. Click this button – the netscreen will then ask if you are sure you want to perform this action. Obviously if you want to reboot it click OK, otherwise click cancel.

Within a few minutes you should be back up and running.

Configure Cisco To Work With Windows NLB

Cisco switch, Microsoft Windows Network Load Balancing.

As part of our new site roll-out we’re using quite a few servers. Some are running in tandem while others are complete backups just in case. Basically, for every function we’re running at least 4 servers (2 load balanced front-facing, 1 backup, and 1 dev/testing).

Using load balancing is a little more tricky on our environment.

On the test environment, which is strictly a vmware server with a crap-load (technical term) of virtual machines, the Microsoft Load Balancing works without any problems. Create a virtual IP, point both servers to it, and away you go.

Unfortunately for us this didn’t work so well on the live production servers. Why? The servers required multicast load balancing. Cisco switches don’t work well with multicast load balancing. The router refuses to learn the ARP for the IP address if it’s coming from a multicast MAC. I had previously switched the servers to unicast, which solved the problem within the Cisco switches, but then the applications would not function on the servers.

One must create a Static ARP Entry on the gateway switch.

For this exercise I have a switch (192.168.1.245) and a server (192.168.1.173) and a client (192.168.2.202) all of which are Class C /24 255.255.255.0 addresses. You will need to know the IP of your server’s Virtual IP (the one for load balancing) and the MAC address of the virtual IP. Easy way to find this is to open a cmd window and type the following:
arp -a 192.168.1.173
OR you can open the network load balance manager and find it listed on the first screen under MAC/physical address.

Now we need to configure your switch

Telnet to your device
Type in your password
Go to enable mode
en
Configuration terminal
config t
arp 192.168.1.173 0100.7f5e.ad01
end
wr mem

Now your pings should work.

Comcast Business with Third-Party Router

I’ve had a few clients that required the use of their current router/firewall combination but wanted the speed of the new Comcast Business Cable Internet connection.

The last client has 16 employees all running on a bonded T1 connection. Maxing out of 1.544mb/sec is no way to live in today’s age. So this client ordered Comcast high-speed to replace the aging T1 (at less than half the cost too), and wanted help getting everything up and running. Why not use the Comcast/SMC router/firewall combination and call it a day? Well, that would be the easy way out. The SMC device is pretty potent for average use, but does not have a VPN server built in. The current configuration has VPN in the mix.

Unfortunately there is no way to setup the SMC into bridged mode, so that makes it a little more difficult to setup. Here’s a little fix I found out after searching for a few hours (yes, hours).

Log into the SMC Firewall (cusadmin/highspeed by default)
Click on the firewall setting
Make sure Disable Firewall for True Static IP Subnet Only is enabled
Make sure Smart Packet Detection is disabled
Check your network settings AND WRITE THESE DOWN (we want the Public IP address which is not a 10.*.*.*, the netmask 255.255.255.252 usually, the gateway which is usually a single IP off from the public IP, and the DNS servers)
Save all settings

Log into your existing router/firewall
Set the IP address of the WAN to the Public IP of the SMC Firewall you wrote down
Set the Gateway, Subnetmask, and DNS entries also to what you’ve written down.
Save all settings

Plug a cable from the SMC Firewall switch to the WAN port of your existing router/firewall. Check your connection by pinging 4.2.2.2

If you’ve set everything up correctly you should get responses by 4.2.2.2 UNLESS you have a rule specifically denying ICMP replies. In that case, just open a web browser window and start running on the internet. You may want to use www.speakeasy.net/speedtest to run a speed test.

After connecting, the Speedtest indicated a connection of 21395kbps down and 8947kbps up. That’s 2674KB (2.6MB) down and 1118KB (1.1MB) up. Not bad at all.

VHCS On Ubuntu

I was given the project of DNS entries for several customers. After playing around with BIND9 for several hours (I actually got it to work for all sites EXCEPT http://sitename.tld – it worked fine for subdomains), a colleague suggested that I try VHCS. VHCS is a free software suite that allows for Virtual Hosts, DNS, and other web related items to be shared and easily managed between several groups. You can grab more data here: VHCS.net

I also chose to put this on Ubuntu because debian packages are easy to install, Ubuntu is fully supported by a huge user base, and Ubuntu also uses a very small footprint.

Once you download the suite, follow the Install directions all the way until you get an error similar to this:
If specified by -literal_key, then the key length must be equal to the chosen cipher's key length of 56 bytes at /var/www/vhcs2/engine/setup/../vhcs2_common_code.pl line 1443
Compilation failed in require at ./vhcs2-setup line (line whatever)
.

Here’s the fix:
nano /var/www/vhcs2/engine/vhcs2_common_code.pl
Control + W and search for db_pass_key
Any time you see 'key' => $main::db_pass_key fill in the following ABOVE the line:
'keysize' => 32,

Then rerun the script ./vhcs2-setup

P.S. By the way even, the instructions don’t really mention that you have to copy some folders over. Make sure you copy everything in the vhcs2-2.4.7.1/configs folder to /etc/vhcs2/ otherwise it will fail out. If you get some postfix errors, that’s probably ok.

P.S. Part 2 There is a much easier product to setup: ISPConfig. You can find it at ISPConfig.com. I recommend using ISPConfig as it is actively maintained.