Ransomware attacks

1,312 Views | 18 Replies | Last: 1 mo ago by IrishAg
eric76
How long do you want to ignore this user?
AG
It is being reported that ransomware gangs are targeting backups more and more. After all, if you have good backups, why pay the ransom?

The big mistake is people thinking that having a copy of the files on another hard drive constitutes a backup. It doesn't.

From https://www.bleepingcomputer.com/news/security/ransomware-has-a-new-target-is-your-backup-ready/
Quote:

Ransomware is far less frightening when you know you can recover.

That's why threat actors are turning their attention to encrypting backups. They wipe recovery points and disrupt the infrastructure needed to restore the encrypted data. Once that recovery path disappears, the pressure to pay rises sharply.

For IT leaders, the lesson is uncomfortable but simple. A backup is only a safety net if the attacker cannot reach it.

...

In February 2024, the ALPHV/BlackCat ransomware group encrypted Change Healthcare's systems after breaking through a remote access portal with no multi-factor authentication. The backups were not isolated or robust enough to restore operations quickly.

UnitedHealth paid $22 million in ransom and still did not get its data back. Total recovery costs hit an estimated $1.6 billion.

...

The BlackMatter group made backup destruction their standard operating procedure. When they hit NEW Cooperative, an Iowa-based farm services provider, and Crystal Valley, a Minnesota farm cooperative, in 2021, they used compromised admin credentials to locate every backup data store and appliance on the network then wiped or reformatted them before encrypting everything else. Since the backups were on the same network, they were easy targets.

...

Gunra ransomware, documented in a joint CISA and FBI advisory in August 2026, pushed that logic further. In one confirmed case, attackers deleted backup and archived data at both the organization's primary data center and its disaster recovery site.

...

Immutable storage as a baseline: Write-once backup copies that cannot be altered or deleted even by someone with admin credentials fundamentally change the attacker's calculation.

...

True network and credential isolation: A backup environment sharing network access and credentials with production is not meaningfully separate.


Too many people think that just keeping a copy of the files on another drive or another computer is a backup. It isn't.

I burn my most important data to CD and DVD disks and have a file cabinet stacked full to the brim with backups.
Lathspell
How long do you want to ignore this user?
AG
Many ransomware attacks also sit on your machine for a while and they let their ransomware get into your backups before locking everything.

For enterprises, "cybersecurity" isn't something you purchase. It is full framework you must implement and practice, in your business. Funny thing is older business owners are generally the most susceptible to fishing and ransomware attacks because they have zero cybersecurity awareness.
eric76
How long do you want to ignore this user?
AG
Most people just want simple solutions that allows them not to think.
AgCMT
How long do you want to ignore this user?
AG
I speak with 100's of SMB customers weekly on their cybersecurity stack. It's amazing how many are stuck in the same setup they have had in place for the past 20 years. I mean, how many iPhones have they bought in that same time frame?

The threat landscape is going nuts with AI. It's just an easier path to entry for the bad actors out there.

If I had a business of any size, especially a small business, I would hire it all out to an MSSP that has a 24/7 SOC. Most of those guys stay on top of what's new in the market. They usually have some sort of training as well.

Over 90% of the attacks are going through SSL VPN's right now. It's hard to convince someone to pay for a ZTNA solution when they have a free or prepaid SSL VPN solution for remote connectivity...but it's so vulnerable.

Featured Video Previews
Lathspell
How long do you want to ignore this user?
AG
The sheer number of small medical practices that don't have an MDR solution in place, email archiving/recovery in place, or even a basic firewall between Comcast and their network is absolutely astounding. We're talking bare minimum...
AgCMT
How long do you want to ignore this user?
AG
Lathspell said:

The sheer number of small medical practices that don't have an MDR solution in place, email archiving/recovery in place, or even a basic firewall between Comcast and their network is absolutely astounding. We're talking bare minimum...


Wow! That doesn't bode well for any HIPA data they are freely handing out.
Lathspell
How long do you want to ignore this user?
AG
The thing is, no one is getting audited for HIPAA. It matters when they become compromised, sure, but it's not easy to convince many of these older doctors that they need to spend an extra $200 or so per month strengthen their situation.
eric76
How long do you want to ignore this user?
AG
AgCMT said:

I speak with 100's of SMB customers weekly on their cybersecurity stack. It's amazing how many are stuck in the same setup they have had in place for the past 20 years. I mean, how many iPhones have they bought in that same time frame?

The threat landscape is going nuts with AI. It's just an easier path to entry for the bad actors out there.

If I had a business of any size, especially a small business, I would hire it all out to an MSSP that has a 24/7 SOC. Most of those guys stay on top of what's new in the market. They usually have some sort of training as well.

Over 90% of the attacks are going through SSL VPN's right now. It's hard to convince someone to pay for a ZTNA solution when they have a free or prepaid SSL VPN solution for remote connectivity...but it's so vulnerable.

I talked to a managed service provider a couple of months ago. While they would have loved to have us as a customer, there was absolutely nothing that they could do for us. For them to even connect to our network would require us to weaken our security.
AgCMT
How long do you want to ignore this user?
AG
eric76 said:

AgCMT said:

I speak with 100's of SMB customers weekly on their cybersecurity stack. It's amazing how many are stuck in the same setup they have had in place for the past 20 years. I mean, how many iPhones have they bought in that same time frame?

The threat landscape is going nuts with AI. It's just an easier path to entry for the bad actors out there.

If I had a business of any size, especially a small business, I would hire it all out to an MSSP that has a 24/7 SOC. Most of those guys stay on top of what's new in the market. They usually have some sort of training as well.

Over 90% of the attacks are going through SSL VPN's right now. It's hard to convince someone to pay for a ZTNA solution when they have a free or prepaid SSL VPN solution for remote connectivity...but it's so vulnerable.

I talked to a managed service provider a couple of months ago. While they would have loved to have us as a customer, there was absolutely nothing that they could do for us. For them to even connect to our network would require us to weaken our security.


Sounds like you have a solid stack already. An MSSP is different than a MSP. There are good ones and bad ones out there. I work for a manufacturer so I see all shapes and sizes of them. If you are really looking for an MSSP I would definitely look at other options. I can give you some recs offline, but they should be able to connect through various means without weakening your security.

eric76
How long do you want to ignore this user?
AG
AgCMT said:

eric76 said:

AgCMT said:

I speak with 100's of SMB customers weekly on their cybersecurity stack. It's amazing how many are stuck in the same setup they have had in place for the past 20 years. I mean, how many iPhones have they bought in that same time frame?

The threat landscape is going nuts with AI. It's just an easier path to entry for the bad actors out there.

If I had a business of any size, especially a small business, I would hire it all out to an MSSP that has a 24/7 SOC. Most of those guys stay on top of what's new in the market. They usually have some sort of training as well.

Over 90% of the attacks are going through SSL VPN's right now. It's hard to convince someone to pay for a ZTNA solution when they have a free or prepaid SSL VPN solution for remote connectivity...but it's so vulnerable.

I talked to a managed service provider a couple of months ago. While they would have loved to have us as a customer, there was absolutely nothing that they could do for us. For them to even connect to our network would require us to weaken our security.


Sounds like you have a solid stack already. An MSSP is different than a MSP. There are good ones and bad ones out there. I work for a manufacturer so I see all shapes and sizes of them. If you are really looking for an MSSP I would definitely look at other options. I can give you some recs offline, but they should be able to connect through various means without weakening your security.

All connections use ssh and come through a bastion host. Both the bastion host and whatever computer they connect to require three public keys to connect -- passwords are only used from the console. And connections to the bastion host are whitelisted.

And for passwords, a minimum password length of 20 characters is required.

Ever since I switched to this approach, we have not seen even one unauthorized attempt to connect.

Prior to that we required IPv6 for ssh connections without whitelisting and we would see one or two connections per month, primarily from companies doing security scans.

I have to run off for a while. I'll add to this when I get back.

I'm back.

For other things like web pages, our pages don't change often so they are static. They are all marked as immutable. Also, the web software runs with its own root (chroot). If something would normally get into the programs and modify them, it won't work because that is all immutable, too. The only thing that is not immutable is the log files and that is coming.

Many of our system log files are flagged with sappnd. syslog can append to them, but the files cannot be modified or deleted except by dropping to single user mode to remove the flags. So if someone did manage to get in, important logs like authlog, daemon, and messages are immutable (except for appends), and so they could not erase their tracks.

I'm thinking of switching to a two firewall system. The outer firewall would have no addresses except for a maintenance network distinct from the LAN and WAN sides. In this way, it could monitor and filter traffic without it even appearing to be there to an attacker.

About fifteen or twenty years ago, I made most system files system immutable. It worked great except that to apply patches, I'd have to drop it to single user mode, remove the flags, apply the patches, and then reapply the system immutable flags. That got kind of old and so I don't do that any more.

Our internal network has one Windows computer and one Mac OS computer. Everything else is one or another flavor of Linux or OpenBSD. Our servers are all OpenBSD.

Our real problem is users who are unconcerned about attacks. I don't think that an MSP or an MSSP is going to change that.
Average Joe
How long do you want to ignore this user?
AG
We have it a little lucky because we can afford to whitelist almost everywhere. Data is also set behind two firewalls, and we use Zscaler for access to on-prem resources. The unfortunate part is that we have yearly HITRUST and SOC 2 audits we deal with, and they suck, so that eliminates a lot of MSSP's. We also have two clients that want to perform their own scans constantly, so constantly having to deal with that.

Similar to eric76, we have jump hosts for everyone. We also block any access that doesn't originate from the U.S (although geofiltering is admittedly easy to get around, it helps with low hanging fruit). We're slowly increasing password size, but we've only made it to 14 characters. Management still pushes back on a big increase, so it's 2 additional characters every few years.

As for backups, we're still kicking it with tapes as our primary backup with secondary cloud backups and warm copies on some slow storage.
eric76
How long do you want to ignore this user?
AG
Quote:

Similar to eric76, we have jump hosts for everyone.

One thing that we do is that the "jump account" is not available for logging into

For example, assume the folllowing:
1) guardian.example.com is our bastion host
2) eric76 is the account on guardian.example.com used for the jump
3) colossus.example.com is the server I want to log into
4) eric76 is the account on guardian.example.com that I want to jump into
5) I am logging in remotely from my laptop, nomad.example.com from a remote location.

Then the following command:
Quote:

ssh eric76@colossus.example.com

will fail because remote logins are unavailable to any account on colossus.example.com, and
Quote:

ssh eric76@guardian.example.com

will fail because it has /sbin/nologin for the login shell instead of something like /bin/ksh and so you cannot log into the account.

In spite of that,
Quote:

ssh -J eric76@guardian.example.com eric76@colossus.example.com

will succeed because we are connecting via eric76@guardian.example.com and not logging into it.

In addition, the ssh keys used for the login to eric76@colossus.example.com are the ssh keys from eric76@nomad.example.com, not the ssh keys from eric76@guardian.example.com

Note that these aren't the computer names and accounts in question.

Note that if you enable ExposeUserAuth in /etc/ssh/sshd_config, it writes your authentication information to /tmp and sets the particular file name in the SSH_USER_AUTH environment variable. You can then use
Quote:

cat $SSH_USER_AUTH

to view the public keys used in the login.

For example, on my actual network, logging into an account on an OpenBSD virtual machine this way from my workstation, the public keys used in the login were:

Quote:

publickey ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAINdHmw55BcuNenYjrA3HMyQl996rPcOMhuTXXn0BSZTO

publickey ssh-rsa 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

and

publickey ecdsa-sha2-nistp521 AAAAE2VjZHNhLXNoYTItbmlzdHA1MjEAAAAIbmlzdHA1MjEAAACFBAEmLm+xF4QU4SY5lqDUxy4/kHzf5ror4dzzZHAXDwESaOzVZ1jrL9jdnsDrMyyM2t5QBmZS9Xs3AXNdLqBqPAT0XwG7r0N4rt6CIWHjgnrQPmJk/0HEzCiZd0mBcvjmNyXENH7yvMR8BXZ8vocfKsFz3ZXvapLwtCMKDscnDl+w7ygdSg==

These are the keys from my workstation, not the keys from eric76@guardian.example.com. The eric76@guardian.example.com account has no need of any ssh keys of its own.

You could take this even further by requiring a separate set of ssh keys used only for the jump account as for the destination account. This would result in needing six unique ssh keys to log into eric76@colossus.example.com from outside the local network.

---

Furthermore, it seems clear that you could have one jump account for everyone to use instead of one jump account for each user. They would generate a separate set of keys for the jump account and provide those to the administrator who could then add them to .ssh/authorized_keys on the jump account.

If the special set of keys are stored in .ssh as id_jump_ed2519, id_jump_rsa, and id_jump_ecdsa, then the /etc/ssh/ssh_config file could contain:

Quote:

Host guardian.example.com
IdentityFile ~/.ssh/id_jump_ed25519
IdentityFile ~/.ssh/id_jump_rsa
IdentityFile ~/.ssh/id_jump_ecdsa

This would then require six ssh keys to log in (a whole lot of overkill) and would require only one jump account instead of many jump accounts.

Disclaimer: I haven't tried using one jump account for multiple users, but don't see why it wouldn't work.

---

In any case, using /sbin/nologin for the jump account would mean that an attacker could not compromise the jump account itself.

---

Regarding my use of a public VPN to protect my servers from ssh attackers has two advantages:

1) A great reduction in the number of entries in /var/log/authlog. If we didn't do this, I would not be able to use "chflags sappnd /var/log/authlog" to make sure that if an attacker was successful, they could not erase the signs of their entry from /var/log/authlog since the /var/log/authlog file size could grow to enormous sizes.

Several years ago, before blocking addresses out of the US, we were seeing roughly 1.3 million log entries in /var/log/authlog from attackers per month per server. Approximately half of those came from 116.31.116.0/24 in China. Ever since then, we have blocked 116.31.116.0/24 from all access to our network and I don't know if they are still active. In a 31 day month with 650,000 log entries per month from 116.31.116.0/24, that would be another log entry ever four seconds just from that one /24 block.

Assume that the average entry is 100 bytes. At 1.3 million log entries per month, that would be 130 megabytes per month per server.

2) More importantly, if a new 0day exploit is found for ssh, whitelisting only specific IPv4 and IPv6 addresses, including those in the VPN, will greatly reduce the number of attackers seeing the ssh services.

---

On a side note, just for grins, I once connected to my account at sdf.org by jumping through three different jump accounts on different servers to reach it.
AustinAg2K
How long do you want to ignore this user?
The problem with good security is that it is in direct conflict with user experience. The more you tighten things down, the more you will inevitably run into users complaining that they now have to jump through hoops. A good CIO will tell those users to kick dirt, but there are bad CIOs who will say, "What's the harm in this one exception..." Then that one exception becomes two, which becomes four, etc.
eric76
How long do you want to ignore this user?
AG
One thing that I am tempted to do is to set up a virtual machine on one of our servers and make that virtual machine my bastion host. It would have only the one account, other than root, That account would have no user shell (/sbin/nologin) and no password and would require three ssh keys to log in. The sole use of the virtual machine would be a bastion host and free up the current for other duties.
AgCMT
How long do you want to ignore this user?
AG
That's a really solid plan! Do you have any end point detection/protection? That's about the only thing I would think to add. It would also identify anything that's already in your system. Something like Sentinel 1 or Crowdstrike.

You know your stuff! I had to read your post about a dozen times to fully grasp it.
Lathspell
How long do you want to ignore this user?
AG
I had a small doctor's office that was already infected with ransomware. Installed SentinelOne, tied into our SOC, and it immediately flagged the threat and the SOC was able to fully remediate.

That was actually one of the only times I've seen the SOC at work because SentinelOne does such a great job at protecting endpoints.
eric76
How long do you want to ignore this user?
AG
AgCMT said:

That's a really solid plan! Do you have any end point detection/protection? That's about the only thing I would think to add. It would also identify anything that's already in your system. Something like Sentinel 1 or Crowdstrike.

You know your stuff! I had to read your post about a dozen times to fully grasp it.

We do a few other things, too.

One thing that we do is block outgoing traffic to known offending IP addresses.

One thing that I used to do was to scan incoming scans to multiple IP addresses and to multiple ports on IP addresses that are not intended for incoming traffic and then block anything and everything from those addresses for several hours. The only reason we don't block them longer is that the blocking table gets too big.

Right now, we add scans of specific ports to a block list, but I will probably go back to monitoring all such traffic again in the not too distant future.

---

My real headache, though, is the users.
AgCMT
How long do you want to ignore this user?
AG
Lathspell said:

I had a small doctor's office that was already infected with ransomware. Installed SentinelOne, tied into our SOC, and it immediately flagged the threat and the SOC was able to fully remediate.

That was actually one of the only times I've seen the SOC at work because SentinelOne does such a great job at protecting endpoints.

S1 is a competitor of ours, but it's not much competition. The are 1a and Crowdstrike is 1b and everything else in the EDR space is a distant 3rd.

IrishAg
How long do you want to ignore this user?
AustinAg2K said:

The problem with good security is that it is in direct conflict with user experience. The more you tighten things down, the more you will inevitably run into users complaining that they now have to jump through hoops. A good CIO will tell those users to kick dirt, but there are bad CIOs who will say, "What's the harm in this one exception..." Then that one exception becomes two, which becomes four, etc.

I call that the executive principle: "your security is great until it stops an executive from accomplishing their task". then it's back to the drawing board because that's no longer a viable solution.

These days I think the two biggest protections for this space are EDR on the front end and proper micro-segmentation on the back end. Being able to control how you see, diagnose, and enforce policies on network flows no matter how the networks are built or where the devices are is becoming a necessity, beyond even ransomware with frontier models deciding to abuse random infrastructure because it might be the most efficient way to accomplish it's task.
Refresh
Page 1 of 1
 
×
subscribe Verify your student status
See Subscription Benefits
Trial only available to users who have never subscribed or participated in a previous trial.