This knowledge base explains known weaknesses in TLS protocols and ciphers, details how cipher negotiation works during opportunistic TLS, and shows how enforced TLS influences encryption standards. It highlights the control customers and sending servers have over TLS use, addresses concerns about weak ciphers, and outlines how Email Security Cloud promotes secure encryption negotiation while balancing security with third-party compatibility.
Email Security.Cloud
1. The mail server clusterX.XX.messagelabs.com still maintains enabled weak TLS Protocols (TLS 1.1, TLS 1.0, SSL 3.0) and possesses activated weak TLS ciphers.
Offered Protocols and Ciphers:
Opportunistic TLS for Mail Forwarding:
Boundary Encryption Service (Enforced TLS):
Customer and Sending Server Control Over TLS 1.0 Usage:
2. Is it possible to disable the weak ciphers?
Diverse Cipher Options in Email Security.cloud:
Cipher Suite Selection by Sending Servers:
Lack of Control Over Sending Servers:
Encouraging Encryption Negotiation:
SMTP Availability in Clear-Text and Encrypted Forms:
Balancing Security and Third-Party Compatibility:
Additional information:
The new enhanced TLS Options and Controls allow customers to specify a specific encryption level for non-enforced connections. With Enforced TLS, customers can choose whether encryption is required on a domain-by-domain basis, and if so, what the minimum TLS version should be. Additionally, a "default" can now be specified, which applies to all other (non-enforced) connections. This means that customers can prevent plain text and weak ciphers for all connections, including opportunistic.
It's important to note that testing this can be tricky, especially when it comes to the Singapore CSA. Since this is shared infrastructure, we don't know who the sender or recipient for the connection will be ahead of time. As a result, we always offer weak ciphers, even if strong ones are configured. To prove that only strong ciphers are allowed, it's necessary to get to the next part of the SMTP conversation (MAIL FROM / RCPT TO). Once we have that information, we will reject the command if the encryption negotiated doesn't meet the level configured by the relevant customers. Testing what ciphers we allow without getting to this part of the conversation is invalid.