Are there any documents available on system hardening for the bigip? Does anyone have any experience in this area?
Thanks in advance.
-m
Are there any documents available on system hardening for the bigip? Does anyone have any experience in this area?
Thanks in advance.
-m
I’d start with a Red Hat hardening guide:
If the system has a few layers of protection in front of it, I’d recommend against tweaking in this way, as most changes will not survive a hotfix or an upgrade.
It would be ideal if F5 would publish a solution or guide to hardening LTM. A few of our enterprise customers have come up with their own methodologies and best practices, but they consider this proprietary information.
Aaron
Yeah, I would offer up my previous work in this space, but it isn’t mine to give unfortunately.
I certainly understand not wanting to release proprietary information.
Posted By hoolio on 07/21/2008 10:06 AM
It would be ideal if F5 would publish a solution or guide to hardening LTM. A few of our enterprise customers have come up with their own methodologies and best practices, but they consider this proprietary information.
It would indeed be helpful if such documents existed.
Thanks,
-m
Been about a year and I’m wondering if perhaps any sort of documentation exists along these lines now?
Thanks,
-m
You might try contacting your F5 account rep and ask them if they have any documents on hardening. If not, you could ask them to escalate the request.
Aaron
Is there any further info about this topic now?
Thanks
mdo
Hi mdo,
I think the same advice applies: you could contact your F5 account rep or F5 Support and ask them to provide hardening advice. This might also make a good request for the documentation forum:
Documentation Requests and Suggestions
http://devcentral.f5.com/Default.aspx?tabid=53&view=topics&forumid=2064
Aaron
Here are my two cents: I think of BigIP as having two very discrete systems: the control plane and the management plane. When most people talk about hardening, they’re talking about hardening the mangement plane (that is, access via the management interface, CLI, etc.).
Changing system settings at this level is generally a Bad Idea, as it’ll potentially break stuff. The best bet is to keep the management network segmented, trusted and secured, and optionally to allow only certain systems CLI or GUI access. As far as the control plane goes - your virtual servers - they’re very secure, at least from the BigIP perspective.
10.1 has incorporated SE Linux, so I expect this will be come less and less of an issue on subsequent releases.
-Matt
Changing system settings at this level is generally a Bad Idea, as it’ll potentially break stuff.
That’s the point of a best practices guide, isn’t it? I understand your position, it just won’t hold water in all environments. All systems in some locations require hardening or they are not certified to be placed on the network. Period.
Version-specific hardening scripts would be handy…any takers? Clearly these will need to be run after hotfixes/upgrades as the changes they enforce are blow away.
My $.02, anyway.
Security is hard. And incovenient. I have no idea what percentage of folks always install the latest feature release, but I doubt it’s too high, so most upgrading would be from existing version to another existing version that would have had time to be vetted. I agree that it’s difficult, I just think it’s a worthy cause. I smell an entrepreneural opportunity for someone. Surely there’s a market for a reseller/partner to tackle this.
Hi - there is a TMOS hardening guide (v2) available which my SE coughed up as soon as I asked. I’d upload it but I can’t see how to upload to this discussion. The main points of it are below;-
- Use “Allow None” on all self-IP as a rule - only allow selected protocols on as as-needs basis.
- Use packet filters for self-IPs on which you have enabled any protocols.
- Disable unused services HOWEVER “It is F5’s considered opinion and recommendation that the security hardening detailed in this guide is sufficient to negate the need to remove non-required services. Note that if services are removed, they will need to be removed again following a software upgrade.”
- You can use hosts.allow and hosts.deny for management interface
- Secure NTPD to work only as a client using ntp.conf
I did feel that there were a few contradictions in the document. At one point the following statement was made;-
…it’s recommend that the management port is not used, access being gained only through the switch ports and the serial console.
then the very next page says;-
Set ‘allow none’ on all Self IPs and only administer the BIG-IP using the Management Port.
Oh well - it at least provides a reference point so that you can say that you have applied all the vendor recommendations.
J
Regarding this:
"it’s recommend that the management port is not used, access being gained only through the switch ports and the serial console. "
The reason some people suggest this is that you cannot restrict access to ports on the management interface. However, there should be a firewall between hosts on the management subnet and any untrusted network or hosts. So I don’t think this is a valid recommendation. And the major advantage of having the management interface available for admin access is that it will still work if TMM doesn’t start. The switch ports will not work if TMM does not start--like if the config doesn’t load or the license is invalid.
So I always recommend having the management port available and protected by an external firewall.
Aaron
This F5 link would help towards hardening - Securing access to the BIG-IP system