A VLAN, or Virtual Area Network, is a list of IP addresses that you control (do they have to be sequential?) and the group of devices using those addresses.
Every AppNexus customer has at least one VLAN in every datacenter where they have equipment.
Extensive use of VLANs provides segregation of each customer's traffic from AppNexus traffic and other customers' traffic. Bi-directional ACLs are applied on every routing interface with a Default Deny policy, meaning only explicitly permitted traffic is allowed to pass. The use of TCP-based protocols provides connection reliability and allows for session protection via ACLs and host firewalling.
Relevant here? There is extensive use of encryption (SSH, SSL-VPN) throughout the network.
VLANs and datacenters
Customers are assigned a separate VLAN in each datacenter where they have equipment. with 8, 24, 56 or 120 IP addresses (how many are used for equipment already?). By default, all ports between your two VLANs will be open. Traffic between LAX1 and NYM1 travels over the Internet and is not encrypted. Customers can request multiple VLANs in a single datacenter if desired.
Assignment of new VLANs could take up to workday, as it involves changes in ACL rules for all your VLANs.
Assigning IP addresses.
Firewall and ACLs
The firewall's ACLs (Access Control Lists) are controlled by the customer. ACL refers to a list IP addresses, both origin and destination, and ports where traffic is permitted to pass. The *router* does this? Access control lists can generally be configured to control both inbound and outbound traffic, and in this context they are similar to firewalls. How are they different from firewalls, is at all? Is the ACL just the "firewall settings"?
If you run out of IP addresses in your VLAN.
You can either migrate to a larger one or, have a second one. Migration currently requires brief downtime.
We will have an API for ACLs. Also the portal.
Relevant tickets: #4128 /#4625/#3952
5. Will migrating be available through the API in the future?
You can open all traffic between VLANs in each datacenter but will travel over the internet.
What about instructions:
1. When you launch new instance, please use --ip= flag of `manage-instance launch` command to explicitly declare IP address of instance in new address block.
2. Please note new command `manage-instance bundle` implemented in last CLI release. With this command you're able to migrate your instances to new address space right now. You need to create images from every your instance and launch instances from these images in new address space. Please note that making image requires shutting down origianl
Migrating to new addressspace is not mandatory right now - in case if you don't need PTR (reverse DNS) records for your old instances. You may wait till we will implement and release features for more convenient migration.
Please, when you launch new instances, explicitly assign them IP addresses from new IP ranges, to not fail into IP migration process with these new instances in the future. To be able to assign IPs to newly launched instances, you need the most recent version of CLI to be installed on your management instance(s).
By default when we assign second VLAN in the same datacenter, we apply to the new VLAN all general (not host-specific) rules from the old VLAN, and open all traffic between VLANs in the same datacenter. Let us know if we should change these rules.
Meanwhile I'll prepare instructions on how to deal with multi-VLAN environment, how to migrate instances from old IP space to the new one with the existing API/CLI functionality, et cetera.
What happens when you run out of IP addresses?
Stateful vs. non stateful inspection.
These questions are also answered on the wiki FAQ but may be helpful here.
1. Why do you use ACLs instead of a stateful firewall?
Stateful inspection is most useful for protecting outbound traffic, but with hosting, the servers tend to receive traffic instead of initiate it. Also, because we are dealing with an unknown amount of traffic, the ability to scale is very important. Stateful inspection is an expensive task for a device to perform and therefore subject to strict capacity limitations (we're talking sub Gigabit for most firewalls). On the other hand, Cisco routers perform ACL packet filtering at line rate with absolutely no performance hit. So, while stateful inspection is appropriate for small, stable amounts of outbound traffic or for protecting niche pieces of the network, (like e-commerce databases), ACLs are more scalable and efficient for protecting inbound traffic to servers. If a customer still desires a stateful firewall, we can add it for a fee.