Secure reference architectures¶
We recommend using SSL/TLS on both public networks and management networks in TLS proxies and HTTP services. However, if actually deploying SSL/TLS everywhere is too difficult, we recommend evaluating your OpenStack SSL/TLS needs and following one of the architectures discussed here.
The first thing one should do when evaluating their OpenStack SSL/TLS needs is to identify the threats. You can divide these threats into external and internal attacker categories, but the lines tend to get blurred since certain components of OpenStack operate on both the public and management networks.
For publicly facing services, the threats are pretty straightforward. Users will be authenticating against horizon and keystone with their username and password. Users will also be accessing the API endpoints for other services using their keystone tokens. If this network traffic is unencrypted, passwords and tokens can be intercepted by an attacker using a man-in-the-middle attack. The attacker can then use these valid credentials to perform malicious operations. All real deployments should be using SSL/TLS to protect publicly facing services.
For services that are deployed on management networks, the threats aren’t so clear due to the bridging of security domains with network security. There is always the chance that an administrator with access to the management network decides to do something malicious. SSL/TLS isn’t going to help in this situation if the attacker is allowed to access the private key. Not everyone on the management network would be allowed to access the private key of course, so there is still value in using SSL/TLS to protect yourself from internal attackers. Even if everyone that is allowed to access your management network is 100% trusted, there is still a threat that an unauthorized user gains access to your internal network by exploiting a misconfiguration or software vulnerability. One must keep in mind that you have users running their own code on instances in the OpenStack Compute nodes, which are deployed on the management network. If a vulnerability allows them to break out of the hypervisor, they will have access to your management network. Using SSL/TLS on the management network can minimize the damage that an attacker can cause.
SSL/TLS proxy in front¶
It is generally accepted that it is best to encrypt sensitive data as early as possible and decrypt it as late as possible. Despite this best practice, it seems that it’s common to use a SSL/TLS proxy in front of the OpenStack services and use clear communication afterwards as shown below:
Some of the concerns with the use of SSL/TLS proxies as pictured above:
Native SSL/TLS in OpenStack services does not perform/scale as well as SSL proxies (particularly for Python implementations like Eventlet).
Native SSL/TLS in OpenStack services not as well scrutinized/ audited as more proven solutions.
Native SSL/TLS configuration is difficult (not well documented, tested, or consistent across services).
Privilege separation (OpenStack service processes should not have direct access to private keys used for SSL/TLS).
Traffic inspection needs for load balancing.
All of the above are valid concerns, but none of them prevent SSL/TLS from being used on the management network. Let’s consider the next deployment model.
SSL/TLS on same physical hosts as API endpoints¶
This is very similar to the SSL/TLS proxy in front but the SSL/TLS proxy is on the same physical system as the API endpoint. The API endpoint would be configured to only listen on the local network interface. All remote communication with the API endpoint would go through the SSL/TLS proxy. With this deployment model, we address a number of the bullet points in SSL/TLS proxy in front A proven SSL implementation that performs well would be used. The same SSL proxy software would be used for all services, so SSL configuration for the API endpoints would be consistent. The OpenStack service processes would not have direct access to the private keys used for SSL/TLS, as you would run the SSL proxies as a different user and restrict access using permissions (and additionally mandatory access controls using something like SELinux). We would ideally have the API endpoints listen on a Unix socket such that we could restrict access to it using permissions and mandatory access controls as well. Unfortunately, this does not seem to work currently in Eventlet from our testing. It is a good future development goal.
SSL/TLS over load balancer¶
What about high availability or load balanced deployments that need to inspect traffic? The previous deployment model (SSL/TLS on same physical hosts as API endpoints) would not allow for deep packet inspection since the traffic is encrypted. If the traffic only needs to be inspected for basic routing purposes, it might not be necessary for the load balancer to have access to the unencrypted traffic. HAProxy has the ability to extract the SSL/TLS session ID during the handshake, which can then be used to achieve session affinity ( session ID configuration details here ). HAProxy can also use the TLS Server Name Indication (SNI) extension to determine where traffic should be routed to ( SNI configuration details here ). These features likely cover some of the most common load balancer needs. HAProxy would be able to just pass the HTTPS traffic straight through to the API endpoint systems in this case:
Cryptographic separation of external and internal environments¶
What if you want cryptographic separation of your external and internal environments? A public cloud provider would likely want their public facing services (or proxies) to use certificates that are issued by a CA that chains up to a trusted Root CA that is distributed in popular web browser software for SSL/TLS. For the internal services, one might want to instead use their own PKI to issue certificates for SSL/TLS. This cryptographic separation can be accomplished by terminating SSL at the network boundary, then re-encrypting using the internally issued certificates. The traffic will be unencrypted for a brief period on the public facing SSL/TLS proxy, but it will never be transmitted over the network in the clear. The same re-encryption approach that is used to achieve cryptographic separation can also be used if deep packet inspection is really needed on a load balancer. Here is what this deployment model would look like:
As with most things, there are trade-offs. The main trade-off is going to be between security and performance. Encryption has a cost, but so does being hacked. The security and performance requirements are going to be different for every deployment, so how SSL/TLS is used will ultimately be an individual decision.