tptacek is using an appeal to authority given his standing as a security expert on this forum. That's the only reason his comment containing no justification is being taken seriously.
Example, same author: Cryptographic Right Answers.
(Disclaimer: I work with tptacek.)
I'd also like to add that you may want to work on the tone of your comments if you'd like to get enough karma to really participate in the community. Your comment history seems a bit confrontational and argumentative. There are better, friendlier ways to make the same points. I'll defer to the guidelines[0] to explain how you may improve your communication skills.
First of all - it takes less than an hour to set up a WAF, and it doesn't worsen your security position. Not only is it free, it's fast and easy and will demonstrably be better than no security at all.
Saying "don't put up a firewall because it isn't perfect" is stupid. You don't have to be a security guru to get that, which obviously several people have, or they wouldn't all be giving you crap about it.
Securing your application long term is a complex, expensive, time consuming process. Suggesting that spending a couple months to be "serious about security" is somehow better than first taking one hour to do the most bare minimum effort is ridiculous.
If I have a computer on the internet, I put a firewall up. Is it defense in depth? No. But that doesn't mean it won't stop the most basic, most common attacks, and it takes no time and no skill to apply it.
I don't know what weird thing you have against WAFs other than apparently they aren't cool or sophisticated enough, but 1) they remove attack surface, 2) they filter the most common drive-bys, 3) they can be both passive and active, 4) they're free, 5) they're easy, 6) they're quick to deploy.
How is it 2018 and I have to explain to you of all people that the web equivalent of iptables might be a good idea if you have zero security?
They do what you configure them to do. Without knowledge of what's allowed and what's not, they're going to have either false positives or false negatives. And if you have that knowledge, you can fix those issues in the web app rather than hide them with the WAF. A lot of preconfigured WAFs will remove specific matches you don't care about anyway rather than attack surface.
> 2) they filter the most common drive-bys
Common as in known proxy or webapp exploits? If that's the case, why not patch them? (There's a reason to use WAF if you can't patch but that costs time/money to get right and test)
> 3) they can be both passive and active
I'm not even sure what passive WAF means... just logging?
> 4) they're free
Only for very small traffic.
> 5) they're easy 6) they're quick to deploy.
See point 1. If you can't put the right restrictions on the app, why do you think you can put them in WAF? If they're not in the app in the first place, why do you think this way is quicker? It's not a magic box will know your requirements. You need a proper setup or the only thing you achieve is removing obviously silly cgi-bin requests from your log files.
I'm not even against WAFs. Just think the gain from them is been none and minimal if you reduce it to "turn it on, it's cheap and easy". It's really similar to iptables - you can turn it on, but why are you doing it? (On a server anyway) If you have a specific reason/route to protect - great. If not, make sure you bind non-public services to local interfaces and you don't need iptables.
> it takes less than an hour to set up a WAF
Not if you want it to do anything useful. Default WAF rules are mostly useless and come with bypasses -- kinda like XSS filters in browsers, and browsers are in a way better position to judge if XSS is going on or not. You can sort-of get useful WAF rules if you encode most of the invariants of your system in your WAF -- but your app is in a strictly better position to do that sort of validation for you instead.
> and it doesn't worsen your security position.
Yes, except for the shiny new TLS MITM, the pathological parser performance cases, the new attack surface in that WAF itself (hands up if you've seen XSS in a WAF interface) and the continued operational cost of having new infrastructure that sees plaintext... Sure! And then I'm being generous in assuming we're not talking about a virtual appliance or as I like to call them "mystery *nix box you can't patch or audit".
> Securing your application long term is a complex, expensive, time consuming process. Suggesting that spending a couple months to be "serious about security" is somehow better than first taking one hour to do the most bare minimum effort is ridiculous.
Also, unlike WAFs, it's effective.
To be clear: I am not conceding your point about cost. It does not take "months" to identify systemic appsec issues. It also does not take an hour to configure a WAF and have it do something vaguely useful for your application.
Anecdotal evidence: Latacora literally does this systemic-discovery-and-remediation-plan thing for a living as a part of our engagement. "Months" is not our timeline for any kind of systemic appsec issue like SQLi or XSS.
> ... the web equivalent of iptables might be a good idea
Are you really sure you want to stand by that equivalence? How often does one come across iptables bypasses?
Yes. Both require some knowledge of how to use the tool and rules, but both are not difficult for even a novice to pick up quickly. Not to mention both can be done by either a dev or a sysadmin in the cloud right now, neither of them require in-depth knowledge of the code, and the scope of the change is simple enough that it can be tested and applied quickly.
> "Months" is not our timeline for any kind of systemic appsec issue like SQLi or XSS.
The discussion was about a dev or sysadmin fixing all this themselves ("taking appsec seriously" etc). Not just running Burp on their website. After they identify the vulns they have to have meetings to discuss fixes, figure out how to patch them and test that it works, go through QA and then roll into production. A fix may involve several systems and most people's deploy process is painful. I've never seen that take less than three weeks, and six is more reasonable. Can you do it in less time? Of course. Does that actually happen? Not without paying a consultancy $300k to hold your hand for two weeks.
It is completely plausible to get a WAF up and running and have it provide real protection immediately. CloudFlare's is probably the best example of this, considering the extensive add-ons they provide based on the intrusions they've seen. I don't like CloudFlare's TLS MITM either, but tens of thousands of their customers don't seem to mind.
Just like iptables, ModSecurity doesn't do anything at first. You have to actually make some decisions. Does that mean you shouldn't spend the extra 15 minutes to read a manual and enable some protection?
Earlier it was suggested that protecting against Drupal or Wordpress vulns is unnecessary. I disagree! There's no reason not to protect against drive-by vulns. Filters for XSS/SQLi, unnecessary HTTP features, malicious bots and scanners, directory traversal, RCE, etc are well worth preventing whenever possible.
As for false positives, they're not as bad as they used to be. There are thorough guides designed to help you identify and remove them from your site. https://www.netnea.com/cms/apache-tutorial-7_including-modse... https://www.netnea.com/cms/apache-tutorial-8_handling-false-... For a sysadmin this may take more time, just like it would if you were using iptables to secure a black box. A dev can change the filters immediately to suit their app.
But even if you don't do any active filtering, the passive filters can at least alert you that something is happening so you can go fix it. This is so standard that Kubernetes includes it as an option in their Nginx Ingress Controller. This guide can get you up and running on your existing K8s install in 10 minutes: https://karlstoney.com/2018/02/23/nginx-ingress-modsecurity-...
This points out how WAF/NIDS is just a speed bump for a serious attacker.
You may conclude that they don't really 1) remove attack surface 2) Unclear if they really do this, and if they do what value there is 3) See the papers for detail on the differences 4) Zero cost to purchase, but what cost to maintain, run, keep up? 5) see #4 6) even if you have a large web farm?
This paper was writen 20 years ago.
Or you go to the doctor and expect him to justify your treatment down to the minor detail? You sound like you talk to your doctor about $overpriced_drug.
It's just reasonable to ask for rational arguments. That's the whole point of the rants here.