0

As a SaaS hosting, we are considering allowing customer to use an alternative domain to access his SaaS environment.

Default domain owned and hosted by us:

customersubdomain.ourdomain.com

Additionally we could add bindings for any domain to the same site, allowing an alternative domain, not owned by us:

www.customerdomain.com

Are there any security risks of doing this?

Improve this question
1

No, if the application is used solely by the customer, or any third party trust the application because they trust your customer. (not you)

For example: A administration panel, a web site that is under the customer's solely Control, a shopping page for customers etc.

Yes, if the application provides value to a third-party, and this third-party should not need to trust the customer. For example if a user select to trust this SaaS application without implication of which customer it is. Another example is a credit card entry form by a PCI DSS compliant aquirer. If the aquirer allows the customer to set his own domain on the CC form, then the customer's DNS server must be protected on the PCI DSS level, since if this are modified, the domain can be aimed on a CC collecting malicious site.

One example of this is that you provide antivirus scanning, and then you sell this "without brand name" to a customer, which can put his own name on the antivirus solution. If the customer trust this solution because the Engine come from you, then your customer can turn malicious and aim the SaaS domain on a malicious site at any time.

Improve this answer
4
  • I must say honestly I have some difficulties following your explanation. Here is what I got out of it: Risk1: DNS related security risk (didn't fully understand this one) Risk2: Customer can read HTTPS traffic of 3rd party accessing the application by their domain Risk3: Customer can modify application content, between our sever and 3rd party, like MITM.. – Slava Feb 9 '15 at 10:55
  • R1: Its hard to explain, but lets say you have a webapplication, that is made by you. Just so you understand, lets call the webapplication TrueCrypt. People trust the web application because its TrueCrypt by you, not because you resell the solution to Customer01. If you resell the solution and allow Customer01 to have their own domain connected to the application, people will use the Customer01's domain to access the application. If Customer01 turns malicious after this, then the malicious Customer01 can then access your user's data. – sebastian nielsen Feb 10 '15 at 6:40
  • R2: No. Customer can't read HTTPS traffic just becase they use their own domain with the application. But some CA's issue certificates solely based on domain Whois records, so it might be a concern. But it would still require a MITM. R3 same as R2. Note that they cannot read traffic belongning to "other customer's users", only their own users. Note that the victim user of the attack MUST acces the SaaS application by the malicious customer's domain for ANYTHING to happen. – sebastian nielsen Feb 10 '15 at 6:42
  • A good idea can be to restrict so if a SaaS application is access through Customer01's domain, only users belongning to Customer01 will be valid. Not Customer02's users. Then all customers will grow a habit of accessing the SaaS application through their own portal and not anyone else's portal, thus eliminating that Customer01 might read or get hold of data belongning to Customer02. – sebastian nielsen Feb 10 '15 at 6:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.