Slashdot Mirror
Ask Slashdot: Is Running Mission-Critical Servers Without a Firewall Common?
An anonymous reader writes: I do some contract work on the side, and am helping a client set up a new point-of-sale system. For the time being, it's pretty simple: selling products, keeping track of employee time, managing inventory and the like. However, it requires a small network because there are two clients, and one of the clients feeds off of a small SQL Express database from the first. During the setup, the vendor disabled the local firewall, and in a number of emails back and forth since (with me getting more and more aggravated) they went from suggesting that there's no need for a firewall, to outright telling me that's just how they do it and the contract dictates that's how we need to run it. This isn't a tremendous deal today, but with how things are going, odds are there will be e-Commerce worked into it, and probably credit card transactions... which worries the bejesus out of me.
So my question to the Slashdot masses: is this common? In my admittedly limited networking experience, it's been drilled into my head fairly well that not running a firewall is lazy (if not simply negligent), and to open the appropriate ports and call it a day. However, I've seen forum posts here and there with people admitting they run their clients without firewalls, believing that the firewall on their incoming internet connection is good enough, and that their client security will pick up the pieces. I'm curious how many real professionals do this, or if the forum posts I'm seeing (along with the vendor in question) are just a bunch of clowns.
So my question to the Slashdot masses: is this common? In my admittedly limited networking experience, it's been drilled into my head fairly well that not running a firewall is lazy (if not simply negligent), and to open the appropriate ports and call it a day. However, I've seen forum posts here and there with people admitting they run their clients without firewalls, believing that the firewall on their incoming internet connection is good enough, and that their client security will pick up the pieces. I'm curious how many real professionals do this, or if the forum posts I'm seeing (along with the vendor in question) are just a bunch of clowns.
Everything is Fine.
Is stupidity common? Yes. Yes it is.
In my experience, the stupid people tend to get fired eventually. But the mess they leave behind can be tremendous.
Not just because plenty of things are run by stupid people, but also because otherwise smart people can have pretty damned important blind spots. And other IT people have been talked out of it by their clients just like you're letting happen.
Whether it's common or not has no bearing on whether it's a good idea.
The only question you need to ask them is weather they're willing to accept the quantified risks from having exposed systems.
It sounds a little like you're trying to just fling a firewall at the system and improve some sort of objective security metric.
What threats are you risks to mitigate with the firewall? What threats will it help guard against?
They don't come for free, and configuring them don't come for free.
I've set up networks where the server infrastructure itself is on its own segment, so there's no need for firewalls between the servers themselves, but the whole subnet is firewalled by a border router.
A lot depends on how tightly you can lock down a server. On my *nix boxes, I tend to only run daemons with listening ports to the extent absolutely necessary. I have a LAMP server that basically has ports 22, 80 and 443 open, and everything else either shut down or set to listen only on 127.0.0.1. Do I really need to configure iptables?
The world's burning. Moped Jesus spotted on I50. Details at 11.
As soon as they start handling credit card transactions, they will need to conform with PCI standards, which will mandate much much higher levels of protections. There are significant fines associated with non-compliance so you may want to forward them over information about this.
The key is to only ever run the services that are absolutely needed, carefully configure these and keep them up to date. If you follow that advice a firewall is an added level of security but not necessarily needed.
Is a POS vendor used by most Australian newsagents. Their contract not only mandates the lack of a firewall, but a writeable share of the C: drive on the Windows machine acting as a server - with no authentication.
While this is incredibly negligent, the support contract makes the vendor completely liable for any security breach that occurs while honouring their contract requirements.
The problem with this battle is that you're a contract worker. So if reasoning/persuasion doesn't work, then you're only options are to end the contract, or fulfill your obligation.
Keep documentation that shows that you brought up the problem, and were rejected. Bake in language on subsequent contracts that give you an out under these types of scenarios, and move on.
If someone is unwilling to listen to reason, is in a position of power, and there's no laws that they are breaking, then that pretty much gives you all of the information you need to know about your options! Just learn to stop worrying and love the bomb.
If your database is in a trusted network zone, it's fine.
If you have a bunch of assets outside the corporate firewall, you're doing it wrong. These belong behind a DMZ firewall, blocking any ports not strictly necessary, possibly with PNAT and coalescence (i.e. an FTP, Web, and Mail server, natted to the same address, ports 80, 443, 25, 21, and FTP PASV going to different addresses behind that).
Within that DMZ, servers provide whatever services they're going to. MySQL on port 3306 will provide MySQL on port 3306; if you add a local firewall, you will have a firewall that blocks all non-listening ports and leaves port 3306 open, so no difference. If you're worried about ssh, use an IP console card (DRAC, etc.) on a separate subnet, or put the database servers behind another firewall. It is, in fact, common to create trust zones for front-end, application, and database, such that i.e. your Web servers connect through WSGI to a CherryPy application, which connects back to a Database, through a firewall in each step. You can do this with vlans and broken-down subnets, one switch, and one firewall.
You have to consider everything when you design secure network architecture.
Support my political activism on Patreon.
From the sounds of it, you're discussing disabling a software firewall, not an actual hardware firewall. There's a lot of applications which require local firewalls to be disabled - for instance, we disable local firewalls when we're deploying telephony application servers because of vendor requirements. Likewise, some applications require SELinux to be disabled as well. All of our servers are still collectively behind a firewall, and beyond that we have a number of ACLs and centralized authentication controlling them. As for not running a firewall being lazy - firewalls are tools. Sometimes they're the right one, and sometimes they aren't. The only way to tell is experience on when to use each tool (and budget too). The more time you spend with networking, the more you'll come to realize that. But since you're learning, stick to what you've been told until you master it. As Picasso said, "learn the rules like a pro, so you can break them like an artist."
Or bandwidth. Or visitors. Or intern-connections. There's a lot of room for serious damage from a lack of security, and not all of it is data theft.
People using your server as a spambot is bad.
People infecting your sites visitors with malware is bad.
People jumping to a different, more secure system from your server is bad.
We tend to notice the data theft issues most these days, because a lot of companies keep a lot of sensitive data, so a Target credit card hijack is tremendously bad and newsworthy.
But that doesn't mean other classes of security risk don't exist.
Your post is not clear on what you mean by "without a firewall". There are so many places in a typical setup where a firewall could be placed, and yes, it is safe to leave them out in some situations.
For example, your store has a firewall at the internet connection and everything inside is a private ip address. The cash registers run on their own network, firewall'd away from the other computers in the store, with rules to allow for outgoing credit card authorizations and that sort of thing. Does each cash register need a firewall? Probably not, and it might even be a significant expense to maintain updated rules every time the network needs to be reconfigured.
So yeah, it depends on the entire configuration. The tone of your post suggests that the situation is not good though, and of course, it's a lot easier to argue for a firewall these days than against one.
You need to look at the PCI-DSS requirements because this is what dictates the security standards of your network if you are storing credit card information. Specifically PCI-DSS dictates (not your contract) that there needs to be multiple levels of firewalls. Ergo you will need a firewall in front of the web server. You will then need a separate firewall in front of the DB servers. And the preferred setup is a three or more tiered system. Web server with firewall connects to the Application (WCF / web service server) which also has a firewall, which connects to your database server which also has a firewall. Also note that I am referencing hardware firewalls such as a Cisco ASA or a Dell Sonicwall. The servers should also have their own software firewalls enabled whether it's Windows Firewall or Linux IPTables. With that said we are "supposed" to be PCI-DSS compliant and should be for the sake of liability (and doing it the right way). Unfortunately I know many vendors who don't want to spend money on proper setups and run very insecure systems. If you can avoid it don't work for these people and go find a client that has the budget to do things right. PCI-DSS: https://www.pcisecuritystandar... A better option for a cheap client is to not store any customer data and use a tokenized system. Authorize.Net will store all sensitive data for an extra $10/mo and allow you to skirt PCI-DSS regulations. You should still run a firewall though and be as close to PCI-DSS as possible though. http://www.authorize.net/solut...
Sorry to barge in like this.
Oracle does not have issues with firewall. A proper firewall will allow a specific program to monitor a range of ports.
Ex.
Open port 80 system wide.
Open ports 40000-65000 for sqlserver.exe TCP and UDP.
You may have multiple listener processes, it takes a few moments and some research but in the end, you ensure the door is opened only for the ports and processes you want. This blocks the door for ports and processes that may be vulnerable thru bugs.
It's not perfect, nothing is. But it's better than staying opened.
Will you get hit if you don't, not necessarily but what if you do??? How much is your data worth? Restore time and data lost since that last restorable backup? What? You don't have a backup or have not tested your restore recently... (excuse me while I rotfl).
Sorry for the nasty punts, but let's face it, the day you get hit. I will say the same thing as today. Rather you hear it today, it's cheaper for you and if I helped in anyway, I'll be glad to not laugh later. I do go see humour shows, I don't need this for entertainment.
Good luck, and best of chances either way you go.
In my experience, many smaller companies, especially ones who offer a specific one-off product, this is a common attitude. This means they've done no real security testing on their product, or how their product is deployed and managed in a customers environment. I think it stems from a couple of things: 1) They aren't security literate. They know how to code or deploy, but they can't be bothered to learn and implement security. They have enough to worry about as it is, and security isn't one of them. It's nothing less than willful ignorance. 2) Sometimes it's more nefarious. They don't want anything impacting their customer experience. Two factor authentication? Firewalls? Application white-listing? Those things get in the way of a customer using their code they paid for. They will not endorse or support it. More over, if YOU implement, it could violate your warranty and null any SLA's. Read the fine print. Ultimately, the (real professionals) answer is this: Defense in depth. For a small business (assuming 1-2 workstations as you've described), a premise (ISP) router based firewall will suffice, and then host based firewalls for each individual client/server/workstation. Keep AV installed, and signatures up to date. Implement a basic change management procedure, and ensure everything stays patched and up to date. All of those things can be done for relatively low cost and high yield for security return. Heck, just doing those basic things puts you head and shoulders above many peers.
I bring nothing to the table.
There are lots of different risks that must be considered when securing a network or system. In my many years of securiy architecture, I've found it make the most sense to create a risk assessment.
Threat x Vulnerability x Impact = Risk
Once you have defined the risks, you can define the best protection method to reduce each risk.
Application firewalls may not be the best protection method depending on the rest of your network security controls. If you have strong network firewalls and every device that connects to the network must be authenticated (and scanned for viruses) before its given an IP address, an application firewall may not reduce much risk. If it doesn't reduce much risk, it may not be necessary.
In business, security is like insurance. You have to justify how much to spend, based on how it will protect us if something bad happens. Further, you have to make sure that whatever the security control is, it doesn't interfere with what the business needs to function. If the database cannot function with a firewall, a firewall is not the best protection method and other options should be considered (Network Intrusion Prevention systems, Data Protection [encryption/tokenization/hashing], Anti-Virus, File Integrity Monitoring, etc). There are many tools available to security professionals today. A firewall is a good tool, but not the only tool... depending on the situation, it may not even be the right tool.
Get a life, not a lifestyle. - Hikem Bey
After 4 weeks of oracle training, the advice from the oracle trainer was that oracle simply doesn't play well with firewalls. I'm not a DBA (thankfully), but that's from their actual instruction.
I've seen firewalls that simply allow any port on any protocol right on through.
Many PHBs seem to think that merely having a firewall is a panacea for all security issues.
If I hear "but it's behind the firewall!" one more time...
Honestly, determining whether you need a firewall isn't as simple as "yes, always, all the time" or "no you don't need one." You have to know what the firewall is doing, and what security is required. You can set up a firewall, allow all ports to be forwarded through without inspection, and while you have a firewall, it's not helping you. Or you could have a server running a secure OS with only the vital ports opened, without access to anything other than the Internet, in which case a firewall probably isn't doing you a lot of good.
Also, it seems you're talking about a software firewall installed on the server? I wouldn't trust it. If I'm running Internet accessible servers, I generally want separate hardware firewall, and I want to put those servers into a separate DMZ if I can. I might leave the built-in Windows firewall turned on if it's not causing any problems, but if I have to disable it, I don't worry too much about it because I have the hardware firewall.
A properly secured Linux/Unix server should be able to sit directly on the Internet without issues, but you may as well put it behind the hardware firewall if you have the option.
But are we talking about disabling the built-in software firewall on a machine that's only accessible by other computers on the LAN? That's probably fine. You should have some security preventing unauthorized personnel from accessing the LAN, and I would assume the SQL databse it password protected, right?
I guess my bottom line here is this: Since you can't trust a the built-in Windows firewall to actually protect from very much, you shouldn't worry too much about disabling it. Make sure your network is secure without it.
Your trainer was an idiot, not a network admin. Oracle database and the various Oracle apps I've used have no silly issues with firewalls. They may have some apps with issues, but not the main product which uses a single consistent TCP port for connections.
Persistent Volume manager for Kubernetes - https://github.com/dwimsey/openshift-pvmanager
A firewall will not stop most attacks. A firewall has to allow traffic to the services that are permitted (duh), and so that open channel, attacking and exploiting the service is what will allow the attack.
A firewall could not stop that by design.
If an internet facing server is secure correctly, there is no need for a firewall in front of it.
There is however a need for a firewall between the DMZ (which is where this server should be) and the internal network, to prevent access to the internal network in the event the server in the DMZ is compromised.
If you ignore ACs because they are anonymous - you're an idiot.
Exactly. Too many people (both businesses and home users) say "Well, I don't have anything that 'those hackers' would want so why bother with protections?" The thing is, though, you DO have something they want. At the very least, a home user has bandwidth. If a malware author hijacks a computer, he can use it to pump out tons of spam. The user might notice an annoying slowdown but otherwise wouldn't know what was up. In the case of businesses, infecting your customers with malware (due to being hacked) or your site slowing down to a crawl (because it is a spam bot and is spending precious resources spamming people) is a sure method to lose customers. I'd wager that the money "gained" by not doing a proper firewall network is more than lost by the "lost sales" of customers fleeing after the servers have been hacked.
My sci-fi novel, Ghost Thief, is now available from Amazon.com.
Don't be a muppet. Limit the spread of malware on your network as much as possible by only opening things that need to be open, to places they need to be open to. There is ZERO reason, for example (plucked at random to illustrate a point), for your end user PC network being able to directly connect to SMB on your SQL server, for example.
Yes, in theory they need credentials to do that. But why leave it open to anyone who obtains credentials when you can be more pro-active about defending the box?
I run: Windows, OS X, Linux, FreeBSD. Just because you have a hammer, doesn't mean everything is a nail.
That design tells me that you need to put a PCI-compliant hardware firewall between the POS and its associated DB server, and the rest of the internal network. And you also need to have a firewall logger that is actually looked at daily, plus you need to do vulnerability scans both internally and externally. A Windows firewall is not sufficient and won't meet PCI DSS requirements in any event, ever, and isn't going to provide any benefit if the firewall between the POS network and the rest of the store/enterprise is in place.
Any device that processes, carries, or stores ANY credit/debit card data that isn't encrypted *must* be behind a firewall that only permits it to send traffic to specific hosts that are necessary for the functioning of the system, and even then only on the bare minimum number of ports, and almost all inbound traffic is denied as well.
My job requires that I deal with a lot of, to put it politely, vertical market software companies. As in, they're the only game in town for that particular function in the industry I work in. It's extremely common to see stuff like this, and it's usually justified by saying "firewalls won't protect you anyway, so why bother?" I only slightly agree -- in my mind the most important thing is to severely limit the use of admin/root accounts and protect their passwords, since you can shut off any security measures once you're through the door.
Usually, it's just laziness on the part of the vendor. The software is assumed to be running on a closed network with no external access in many cases, and a lot of people don't get that even closed networks aren't really closed anymore. I'm completely platform-agnostic, but I've noticed this a lot with typical Windows DCOM fat client / SQL Server (or worse, Jet/Access DB) pairings. As soon as you try to run these securely on a general purpose desktop, you find that port-based firewalling is very difficult to do without opening a huge range of ports due to the way RPC works. Yes, there are workarounds, but in general the protocol is not firewall-friendly. And, the golden rule of vertical market software is "thou shall not upgrade thy technology stack, ever." I do desktop systems integration -- OMG, getting poorly coded VB6 applications working on Windows 8 is a nightmare even with the compatibility toolkit, etc. Not sure what it is with the market segment I'm in, but I see lots of VB6 married to a Jet database, and lots of craptastic fat Java applications. Both can be killers to fix and get working without access to the code/programmer.
To be fair, just because the network is properly setup and allows for certain behavior etc. does not mean the application will play well with that setup. I've seen it happen before (and have been able to demonstrate it with proprietary software) that sometimes the network will not react correctly with certain network setups. At my company we have had to implement special protocols and features in our software just to overcome some inherent network limitations that our IT group pretty much said, "we have no way around this issue, sorry."
I only have limited experience with Oracle though, so take that statement as what you will. I have seen other software exhibit this kind of behavior though.
they went from suggesting that there's no need for a firewall, to outright telling me that's just how they do it and the contract dictates that's how we need to run it. This isn't a tremendous deal today, but with how things are going, odds are there will be e-Commerce worked into it, and probably credit card transactions... which worries the bejesus out of me.
I suggest you relegate the 'is this common' question to a discussion after hours over a beer.
Your real issue is security. I would want to schedule 2 different meetings, preferably with everyone attending in person. The first is a prep meeting with your client and their legal counsel to discuss your concerns, review the contract language that is being referenced by the vendor, and what liability the vendor has if the machine is compromised due to the vendor required there be no software firewall.
The second meeting would be with your client, their legal counsel, and the vendor.
FYI, oracle requires ports from 1024-65535 to be open for any client. 1521 is only used for initial setup dialog. This also forks a new oracle process, which gets its own socket. Afterwards, as stated above, this information is sent back to the client which reconnects on the new socket. This oracle doc explains what I am talking about: https://asktom.oracle.com/pls/...
My current company has a firewall for the incoming internet connection. (What sane company doesn't?) We also have individual firewalls on each PC but no individual firewall on any server. I'm not a network administrator -- it's a black box from my viewpoint, but I can rattle it and guess what's inside. The servers, I believe, are protected in two ways -- (1) to get out on the internet, you must go through a proxy, and the servers do not know how to do that. (2) traffic on the server subnets are blocked by the internet firewall, except for a few in a designated DMZ. We run into this all the time when applications have features that report back to vendor tech support, but are always blocked by the firewall. (In one case we had an application that would hang when it couldn't make an ftps connection with the vendor's tech support site -- who the heck uses ftps anymore? We stopped using that app.)
So to answer your question, a well designed network will have clients that can get to the outside through a proxy server, and servers that can't get to the outside at all, and servers that can cautiously get to the outside from the DMZ. The servers that are blocked from getting to the outside by the network don't necessarily have to have individual firewalls, and in fact, local firewalls can cause problems with some applications.
Now, if you're running the back end part of the system on a local PC that can also get out on the internet... whoo boy... that sounds dangerous.
Oliver's law of assumed responsibility: If you're seen fixing it, you will be blamed for breaking it.
Unless *all* datafiles on your client's system are encrypted, also, and I don't think even that's enough.
ObDisclosure: I worked for about 4 months on a contract at Trustwave, a root CA.
Leaving that huge hole in your defenses... I suggest you look, if you don't already know, at .
From the 1.2 std: "Firewalls are a key protection mechanism for any computer network. Other system components may provide Firewall functionality, provided they meet the minimum requirements for Firewalls as provided in Requirement"
Even all data between two systems *MUST* be encrypted, for full compliance, if you're doing your own.
So, what this vendor is doing... I'd say you and your client need to reread the contract *VERY* closely, and if they say they're adhering to stds, they're in violation of the contract.
mark
"A firewall is a MUST".... However
1.) If you are a contractor (or an employee for that matter), and you're explicitly told not to do something, document your concerns, in writing and either do what they tell you or quit. This covers your butt when thing blow up. Note I said, when, not if. You don't get to dictate the business objectives in either case (contractor or employee).
2.) The networks/firewalls are only one part of the picture. I've noted other posts asking questions concerning the surrounding environment. These are good, even if they don't address point (1.) If a "best practice" for an element in the environment breaks the environment, then the "best practice" is invalid and MAY be ignored. It may also be that the app in question is brain dead, but that's a whole 'nother topic.
I've seen way too many "best practices" that do not account for the specific environment and even more "checklist" jockeys who can't figure out how to make things work without them. Learn to do more than to follow a flowchart/checklist.
This whole "network security" meme is a failed experiment needing to be called out as the ridiculous farce that it is. Firewalls are the equivalent of mounting castle defense against an airforce. It does not work and never has. The opportunity cost of squandering resources on castle building vs standing up an opposing airforce is both high and sad.
Host-based firewalls are even more amusing. They come enabled by default. High probability any application installed needing to listen() is going to automatically punch a hole to do so as vendors have zero interest in dealing with firewall support nightmares. This begs the obvious.. what is the effective difference between listen() and firewalled-listen() ?
If you really want a secure internal system then lock down services/listeners and configure each system to use only secure communication protocols. If this is not possible set IPsec = required and secure the transport E2E only.
The further away from application domain you apply security the increasingly worthless that security is.
How about this: If you find yourself needing a firewall, your system design has already failed. Every single system should assume actively hostile environment.
I can only repeat your original subject.
There are firewalls and firewalls. If you have a box with port 80 listening and nothing else listening, and you install a firewall that blocks connections to ports that are not port 80, you haven't really added a whole heap of useful. If you have a firewall that allows port 80 from anywhere and ssh connections only from a particular subnet, then maybe the firewall is earning it's keep. If you have a firewall that warns you about outgoing connections from your server that might be to command and control botnet servers if it gets compromised, then maybe it is worth doing. If you have a stateful firewall that is doing traffic analysis and shutting down particular attacks on your server then great. Shutting closed doors just for the sake of buzzword compliance is not a useful function.
I think you are pretty confused. If you need to use enterprise level tools you use enterprise level hardware and network configurations. This means that, if you are going to use it you have a separate NIC for each node and an "Oracle Only" subnet. If you don't / can't do that, you are most likely using a tool for which there is no actual need. In other words, you're doing it wrong. Even in this case, you should certainly be blocking the unneeded ports in the 0 - 1000 range.
Guns don't kill people; Physics kills people! - John Lithgow as Dick Solomon on Third Rock From The Sun
Application support always says to turn off everything that might possibly interfere with their precious application. They would have you shut down the operating system if they didn't need it. Application support lives in a fairy land where the only thing they have to worry about is their application. They don't have to fix anything if the application isn't broken. They have no interest in anything else. A good vendor will program their application to work with the system standards. Most ISVs are not good vendors.
As a system or network admin, you have to protect the application from the rest of the network and protect the rest of the network from the application and protect everything from the users and the Internet. Part of doing that is firewalling the crap out of your core network, and if you can't do that you should be looking at adding more VLANs and controlling traffic that way.
The road to tyranny has always been paved with claims of necessity.
and we don't keep logfiles, so we don't have to worry about checking for breakins and cooptions. hey, we don't comment or document our code, either, it's just us two guys. that way, we get to keep all the millions.
hang on, the phone just started melting and my screens went blue...
if this is supposed to be a new economy, how come they still want my old fashioned money?
I setup all our Oracle databases.. (Many, many of them). Only port 1521 is open in iptables. (actually, for some, I have secondary listeners).. The Junipers also ONLY allow access on port 1521, (and a secondary, if specified)
I have never had issues connecting to the database.
What are we going to do tonight Brain?
Highly uninformative post as it doesn't mention this situation is only for NT. On UNIXes, port 1521 (or whatever port is selected in installer) is enough.
Oracle's emphasis is rarely on Microsoft's operating systems. Not only in RDBMS, but many other products support UNIXes primarily, and Microsoft's operating systems secondarily.
Bingo Dictionary - Pragmatist, n. A myopic idealist.