Cloud

5/25/2017
10:50 AM
Connect Directly
Twitter
LinkedIn
Google+
RSS
E-Mail
50%
50%

82% of Databases Left Unencrypted in Public Cloud

Personal health information and other sensitive data is left exposed as businesses overlook encryption and network security.

The average lifespan of a cloud resource is 127 minutes. Traditional security strategies can't keep up with this rate of change, and 82% of databases in the public cloud are left unencrypted.

These findings come from the RedLock Cloud Security Intelligence (CSI) team's "Cloud Infrastructure Security Trends" report. RedLock today formally announced the CSI team and its inaugural report, which focuses on major vulnerabilities in public cloud environments.

The team analyzed more than one million cloud resources, processing 12 petabytes of network traffic, and dug for flaws in public cloud infrastructure. They found 4.8 million records, including protected health information (PHI) and personally identifiable information (PII), were exposed because best practices like encryption and access control aren't enforced.

"Imagine the day and age we live in," says RedLock cofounder and CEO Varun Badhwar. "You should be using encryption of data at-rest. There is no data out of the reach of bad actors if not secured correctly."

The problem isn't in cloud providers failing to secure data centers, but in organizations failing to secure applications, content, systems, networks, and users that use the cloud infrastructure. "That is where people are not aware, or not investing the right resources," he continues.

Researchers found of the 82% of databases left unencrypted in the public cloud, 31% were accepting inbound connection requests from the internet. More than half (51%) of network traffic in the public cloud is still on the default web port (port 80) for receiving unencrypted traffic. Nearly all (93%) public cloud resources have no outbound firewall rule, says Badhwar.

"You need to have control at the network, configuration, and user layers so it's hard for someone to get in, and harder for them to take your data out," Badhwar emphasizes, adding how weak network controls lead to trouble. "It's like saying, 'I'm going to leave my gates and front doors open, and hope I don’t get robbed,'" he says.

Developers and the team running operations in the cloud need to have secure access, and researchers discovered they often don't.

Businesses are moving to the cloud from on-prem environments where everything underwent a security review and sign-off process before being pushed to production, Badhwar continues. Two hours and 27 minutes, the average lifespan of a cloud resource, is a much smaller window.

"Within that timeframe, the customer has no clue how to get security right because developers are pushing code," says Badhwar. "None of the existing security tools work at the speed of change. Customers have no visibility into the changes pushed to production."

He calls the current cloud environment a "devops-oriented world" in which those who write the code are responsible for pushing it to production. The problem is, those who are making changes within cloud environments are not trained security professionals.

Their lack of expertise brings additional risk, especially with new tech like containers. RedLock researchers found 285 Kubernetes dashboards (web-based admin interfaces) deployed on Google Cloud, Microsoft Azure, and AWS that were not password-protected. There were many cases where Kubernetes systems held plaintext credentials to other critical systems, a vulnerability leaving key infrastructure exposed.

Security recommendations from the report include training developers on security practices for public cloud infrastructure, ensuring services are set to accept internet traffic on an as-needed basis, and setting a default "deny all" outbound firewall policy. You should also automatically discover database and storage resources as they are created in the public cloud, and monitor network traffic to ensure those resources are not directly interacting with internet services.

Related Content

Kelly Sheridan is the Staff Editor at Dark Reading, where she focuses on cybersecurity news and analysis. She is a business technology journalist who previously reported for InformationWeek, where she covered Microsoft, and Insurance & Technology, where she covered financial ... View Full Bio

Comment  | 
Print  | 
More Insights
Comments
Oldest First  |  Newest First  |  Threaded View
DominicP260
50%
50%
DominicP260,
User Rank: Apprentice
5/25/2017 | 12:36:43 PM
This article and the cited report are misleading
This article, and the cited report, suggest that the 82% of unecrpyted databases in the public cloud would suddenly become safe if they were all encrpyed but that's far from the truth. This will only protect the database from an attack vector where the threat actor attempts to steal the DB in its entirety at the filesystem level. If an attack exploids weak DB credentials via an exposed port they will still have full read access to the encrypted DB unless the DB is utililzing field level encryption, which is not even mentioned in this article. The way in which the cited report collected their data is also not very scienctic; the 82% is probably not an accruate reflection of the real number. This article reads more like sponsered content than anything.
ChrisB964
50%
50%
ChrisB964,
User Rank: Apprentice
5/26/2017 | 5:22:41 PM
What exactly is the threat vector of an unencrypted cloud database?
What exactly is the threat vector of an unencrypted cloud database?  Considering this is headline I'm curious if you or anyone could describe a scenario where someone accesses data by attacking an unencrpted database.  Unless I misunderstand something the encryption protects attempts to access the data at the disk level.  For RDS users(the most common case) wouldn't an attacker first have to compromize AWS infrastructure to get at the disk layer of RDS?  There's no public access to the disk of an RDS instance.  Help me understand the risk.

Yes we encrypt our cloud databases and noticed that we have to spend more hourly on many due to the additional size requirements of encrypting them.
Dr.T
50%
50%
Dr.T,
User Rank: Ninja
5/29/2017 | 2:36:17 PM
Not surprised
This is not surprising at all. I bet it is the same on on-premises. We tend to think it is important to encrypt data but not doing that when it comes to data at rest.
Dr.T
50%
50%
Dr.T,
User Rank: Ninja
5/29/2017 | 2:38:40 PM
51% web traffic
51% web traffic being not encrypted is more troubleing than database encryption at rest. So individuals who are registering those sites not having https are competely at risk.
Dr.T
50%
50%
Dr.T,
User Rank: Ninja
5/29/2017 | 2:41:22 PM
cloud providers failing to secure data centers
This is something we can argue but it is still both cloud provers and consumers to encrypt the databases. Cloud providers could make this a default features on their services.
Dr.T
50%
50%
Dr.T,
User Rank: Ninja
5/29/2017 | 2:45:17 PM
devops-oriented world
"devops-oriented world" in which those who write the code are responsible for pushing it to production." I do not have problem with this, therebcna still be enough chekcs to make sure no security vulnerability.
Dr.T
50%
50%
Dr.T,
User Rank: Ninja
5/29/2017 | 2:47:39 PM
plaintext credentials
I was hoping that we would not hear this again, plaintext credential means no credential basically.
ScottBinDC
50%
50%
ScottBinDC,
User Rank: Apprentice
5/30/2017 | 12:40:34 AM
But the cloud is supposed to be more secure!
Isn't that what everyone tells us? The cloud is more secure. Why worry? (sarcasm assumed)
sngs7dan
50%
50%
sngs7dan,
User Rank: Strategist
5/30/2017 | 9:45:46 AM
Lies, Damn lies and Statistics
A shameful article. Inadequate with regards to analysis. Either provide enough information to make the numbers quoted reasonable or critique the methodology of the report for not providing enough information.

"'Average' is a statistical fiction". Under three hours average TTL for a cloud resource? Based on what? People creating something, realizing it's misconfigured, destroying it and then doing it right? Trial and error, experimentation? transient convenience builds? dynamic honeypots? the possibilities are endless and so are the questions left unanswered.

Without any discussion of the distribution curve, a single value is a data point without meaning. Imagine trying to read a graph with unlabeled axes. Pretty picture, but with what meaning?

You can do better.
Meet 'Bro': The Best-Kept Secret of Network Security
Greg Bell, CEO, Corelight,  6/14/2018
Containerized Apps: An 8-Point Security Checklist
Jai Vijayan, Freelance writer,  6/14/2018
Four Faces of Fraud: Identity, 'Fake' Identity, Ransomware & Digital
David Shefter, Chief Technology Officer at Ziften Technologies,  6/14/2018
Register for Dark Reading Newsletters
White Papers
Video
Cartoon
Current Issue
Flash Poll
Twitter Feed
Dark Reading - Bug Report
Bug Report
Enterprise Vulnerabilities
From DHS/US-CERT's National Vulnerability Database
CVE-2018-0291
PUBLISHED: 2018-06-20
A vulnerability in the Simple Network Management Protocol (SNMP) input packet processor of Cisco NX-OS Software could allow an authenticated, remote attacker to cause the SNMP application on an affected device to restart unexpectedly. The vulnerability is due to improper validation of SNMP protocol ...
CVE-2018-0292
PUBLISHED: 2018-06-20
A vulnerability in the Internet Group Management Protocol (IGMP) Snooping feature of Cisco NX-OS Software could allow an unauthenticated, adjacent attacker to execute arbitrary code and gain full control of an affected system. The attacker could also cause an affected system to reload, resulting in ...
CVE-2018-0293
PUBLISHED: 2018-06-20
A vulnerability in role-based access control (RBAC) for Cisco NX-OS Software could allow an authenticated, remote attacker to execute CLI commands that should be restricted for a nonadministrative user. The attacker would have to possess valid user credentials for the device. The vulnerability is du...
CVE-2018-0294
PUBLISHED: 2018-06-20
A vulnerability in the write-erase feature of Cisco FXOS Software and Cisco NX-OS Software could allow an authenticated, local attacker to configure an unauthorized administrator account for an affected device. The vulnerability exists because the affected software does not properly delete sensitive...
CVE-2018-0295
PUBLISHED: 2018-06-20
A vulnerability in the Border Gateway Protocol (BGP) implementation of Cisco NX-OS Software could allow an unauthenticated, remote attacker to cause a denial of service (DoS) condition due to the device unexpectedly reloading. The vulnerability is due to incomplete input validation of the BGP update...