Tuesday, 15 April 2014 14:42

No Heartbleed Here

Written by

While organizations spend the next few days and weeks patching OpenSSL vulnerabilities, the realization is setting in that we may never know the full extent of the damage caused by Heartbleed. What we do know is Gazzang services were not impacted by the bug. 

HeartKeyAlthough Heartbleed was only announced in early April, it has actually been present in OpenSSL versions dating back to March 2012. This means hackers have had ample time to steal certificates and other sensitive information. Making matters worse, it’s nearly impossible for companies to know whether their web communications have indeed been compromised.

Should I worry about my Gazzang zNcrypt keys being exposed?

No. Gazzang zNcrypt keys are encrypted client-side, so a compromise of the zTrustee server using Heartbleed would never expose any zNcrypt keys. Furthermore, while we use SSL for data-in-transit encryption, the payload of data between client nodes and zTrustee is encrypted with strong crypto libraries like GPG underneath OpenSSL. So we’re doubling up the encryption, just for instances like this.

Like many other websites, we have already patched our zTrustee SaaS servers for the Heartbleed vulnerability. We also encourage customers who haven’t already done so to upgrade to the latest operating system version and deploy those OS patches as well.

What exactly is being exposed?

When exploited by a hack, Heartbeat (the name of the transport layer security extension where the bug was found) dumps whatever data might reside in the memory of client/server communications in small 64k chunks. Normally this traffic is encrypted, but the bug actually compromises the secret keys, usernames and passwords that protect this data. Leaked keys can lead to insecure web certificates, which could indirectly lead an attacker to usernames and passwords, payment card details, cookies -- essentially any information submitted by other users of the service.

How can I protect my organization against future threats like Heartbleed?

One of the reasons this bug is so widespread is because it exploited a vulnerability in the popular and highly regarded OpenSSL crypto library. In other words, it went after the very service layer that untold numbers of companies use to protect against hackers. Where many of these companies went wrong is they relied on that single layer of security to protect against a network attack.

Multi-factor authentication, which requires a second piece of information to allow access to an account, is one way users can protect email access and other sensitive account information. So in addition to upgrading, patching and maintaining the latest versions of your OS and software, another way to protect your company’s data is to deploy multiple layers of cryptography.

I mentioned earlier that we use GPG in addition to SSL for data-in-transit encryption. As another example, our customers use Gazzang zNcrypt to encrypt their data and protect that data by disallowing unauthorized people and processes to access it. The encryption key is then encrypted itself and stored in the zTrustee key manager (along with the master). The data owner can then set a broad range of configurable policies governing who or what can access those keys.

The important thing to remember is that security needs to be applied in layers, and a single layer is never enough. A useful tool to check your SaaS vendors’ security is Qualsys SSL Labs test.

What can I do as a consumer?

To start, here are a couple of lists spotlighting companies that use the TLS Heartbeat extension. The best advice is to change your password if a service you use is listed as vulnerable.

Monday, 07 April 2014 11:24

Performance tuning Cassandra on AWS

Written by

At a recent San Francisco Cassandra Meetup, we talked about speed for over an hour. But you can get the gist in about 10 minutes by grabbing the deck I presented. While it may be fun and games to talk about speed when it relates to motorcycles, super cars or jet planes, when you are talking about performance tuning your Cassandra cluster it is not a trivial task. You need to tune at all the layers, the DataStax layer, the AWS layer and the core filesystem layer.

We started the discussion with the AWS configurations we’ve seen perform best in real life for running Cassandra. There are a LOT of choices you can make when spinning up AMIs for any cluster, and I shared the OS and hardware choices we’d recommend: Amazon i2 instances have I/O performance that is off the charts.EddieatMeetup

We also covered tuning of Cassandra itself, and of tuning the file system. What steps you take - in what order - matter a lot when you’re trying to capture every ounce of speed. Always baseline before you tune, then tune, then test after tuning. Make sure your results are repeatable at least 3 times before making business decisions based on your data.

The audience seemed particularly interested in the AWS tools you can use to benchmark raw file system I/O and to stress test in the AWS environment. Because I’ve seen these contaminate test results, I highly recommend you note factors like time of day and when the cache was last cleared before starting any test.

And - of course - I presented stress testing results for reading and writing encrypted vs. unencrypted data, running on both Elastic Block Storage (EBS) and Solid State Disk (SSD) storage. The net of this was that data security and high performance can absolutely co-exist, even and especially in the Amazon public cloud. Check out the test results yourself in my deck, and those results can be tuned even further. Feel free to share with any skeptics at your organization who are arguing to shortcut data security on the false premise of a huge performance hit.

BONUS for you:

Don’t have time to do a lot of tuning from scratch? In the spirit of open source code, I’ve posted on github a performance-tuning script we built specifically for Cassandra running on Amazon Linux. Even if that’s not your exact environment, I encourage you to grab it and tweak it for your own needs. Why reinvent the wheel, right?

The benefits can be downright game changing. With your infrastructure hosted in a public cloud, your technical personnel can be redeployed to focus on supporting your business objectives vs. maintaining and optimizing a server farm. You’ll see a nice reduction in CAPEX as well.

And because public cloud resources can be provisioned on demand, your can easily add more compute capacity when you need it most. Think a retailer during the holidays or an accounting firm in early April.

Businesses that are agile and efficient can respond to the marketplace quicker, at lower operating costs, and with more resources deployed to meet customer needs. But according to 451 Research's recent survey (Hosting and Cloud Study 2014), only 7 percent of organizations in North America defaults to the cloud for new application development. So why don’t more companies leverage the compute power of the cloud?

Well, for starters, since your data and applications now exist outside of your office, there’s always a concern about where the data is and who can see it. Does it meet compliance? Can it be compromised? Over 25 percent of the 451 Research survey respondents note that security issues is the number one reason to hold off on cloud adoption.

These concerns are valid and that's why you should make sure that when you look to the cloud, you look for a security solution that provides not only data encryption, but allows you to manage your own encryption keys - something Gazzang has built its security suite around.

Gazzang CloudEncrypt enables Amazon Web Services (AWS) cloud customers to realize the benefits of the public cloud with a portfolio of encrypted machine images. That means when you “spin up” a cloud image, it’s fully encrypted at boot, the encryption keys are secure and in your control, and a number of other security best practices and settings are preconfigured. It’s all done for you, so you can spend more time working and less time worrying about which ports to open and close.

And CloudEncrypt is elastic, just like AWS, so as when you need more compute resources, just provision more CloudEncrypt instances through the AWS Management console. Each new image comes with the exact same security configuration making it easy to grow your secure cloud.

Suddenly those clouds don’t look so threatening, do they?

Few companies are enjoying a better run of news right now than Cloudera. In mid-March the big data bell cow announced $160 million in funding led by T. Rowe Price. Less than two weeks later, Intel’s mega investment of $740 million is still a popular topic around our company’s water cooler (yes, we have a water cooler). 

The company’s latest salvo happened this morning while most of the west coast was still asleep. Today Cloudera announced the general availability of Cloudera 5, the solution that will drive what Cloudera refers to as the enterprise data hub. In short, the hub is a centralized platform where companies can store, process, and analyze all of their data and run any variety of projects. The idea being to make it easier to store everything and then use the data when they need it. 

Cloudera and Gazzang have a longstanding partnership with several mutual customers including Kaiser Permanente and Western Union. We are pleased to be able to announce our foundational zNcrypt and zTrustee encryption and key management solutions are now C5 certified. In addition, Gazzang is one of only a handful of Cloudera partners that have a parcel available for customers to download through Cloudera Manager, so installation is fast and easy regardless of the size of the environment. That means whether your C5 deployment is 10 nodes or 10,000 nodes, each encrypted node is as easy to spin up as the next, and all communicate seamlessly with our software-based key manager. 

The bottom line is companies that must meet a compliance requirement like HIPAA or PCI-DSS - or have some other obligation to protect sensitive data - can continue to feel confident that the business-critical information resident in their enterprise data hub is secure at rest and protected against unauthorized access or attack. 

Beyond certification and automated deployment, we’re also watching the Intel investment with great interest. I’m not going to speculate on what this investment means for either company. Plenty has already been written about it. What’s undeniable though is that software that integrates with or runs on Cloudera now should also be optimized to take advantage of Intel hardware. The good news for customers is we're already ahead of the game.

Gazzang’s big data encryption solution, zNcrypt, was designed to leverage the Intel AES-NI encryption instruction set that can be found on most Intel Xeon and Core i7 processors. We’ve done extensive testing, and when running Gazzang in a well configured Hadoop environment on Intel hardware, customers often see the performance impact of encryption dip into the low single digits on a percentage basis. Check out our Hadoop performance guide to learn more.

Gazzang also leverages Intel technology to generate strong encryption keys. As you know, data encryption really only works if your keys are well protected and separated from the encrypted data. Equally important to how you store your keys, is how they’re generated. A strong key requires good random numbers. The greater the randomness the harder the key is to break. Our encryption solutions leverage the Intel RDRAND Instruction set, Intel’s digital random number generation hardware, to create powerful 256-bit keys that our customers rely on to protect their most sensitive data. 

Together with Cloudera and Intel, Gazzang is able to deliver enterprise big data and cloud security that installs in minutes, runs at peak performance and protects your most important business asset… your data. 

Tuesday, 04 March 2014 15:04

Talking Data Privacy at SXSW

Written by

This weekend, I'm hosting a core conversation session at SXSW, titled, "Dear Taco Vendor, how are you securing my data?" When I submitted the topic, I thought the session would generate some good conversation, and maybe even make some people think. MAYBE. Mostly though, I loved the clever title (kudos to my wife for coming up with it) that combines one of my favorite foods with one of my favorite topics. 

The gist of the session was this. Do you really know what you're getting when you trade your email address, scan your phone or provide any other type of personal information in exchange for free stuff? Where does this data go and how is it secured? Is it at risk for theft?

I work at a cyber-security company, so I'm not naive to the fact that there are certain dangers that come as a result of the wonderfully ubiquitous "series of tubes" that is the Internet. At Gazzang, we often deal in hackers, rogue employees, and vulnerabilities in modern data architectures like NoSQL and Hadoop. Our goal is to help companies keep sensitive data from being exposed. But In researching my session topic, I was amazed at how easy it is to expose someone's very personal identity simply by having access to their email address.  

Toss a few bucks to a data aggregator, and there's almost nothing you can't find online. For example, a quick search of my gmail address turned up my birthdate, last four residences with property values, the names of all my closest relatives, a ton of photos, my work history and links to pretty much everything I've said or done on social networks. 

So much for an email address not constituting personally identifiable information. 

My SXSW session isn't going to focus on whether shady people can access your sensitive data simply by knowing your email address. It's clear that they can. Instead, I want to focus on what that revelation means in a broader context: 

  • Do we need to reset our expectations on privacy, or is that a defeatist attitude?
  • When you give any information to a 3rd party, what should their obligations be to keep it private?
  • How can the public influence vendors to change the way they store and exchange data?
  • What needs to happen (if anything) to change public behaviors around freely sharing sensitive data?

Also, we can talk about tacos.

I hope you'll join me this Saturday at 3:30pm at the Sheraton.

Data-at-rest encryption is essential. It's a requirement for meeting compliance regulations like HIPAA, PCI, SOX and FERPA and is one of the most effective methods for protecting sensitive, business-critical information. What you may not realize is that - in addition to providing an "insurance policy" against data theft - encryption can also be an important revenue driver. More on that in a moment.

IBM Informix customers now realize the benefits of data encryption and key management from Gazzang, the leader in big data security. Gazzang's software suite has recently completed Informix certification. That means users of the hybrid database system can secure sensitive SQL and NoSQL data at rest with near zero performance impact to disk i/o or CPU utilization. 

Gazzang does not require users to modify their Informix database nor the applications above it, and the encryption can be deployed on each datanode within minutes using standard DevOps scripts from Chef and Puppet. The solution supports a range of database types including SQL and MongoDB and currently encrypts more NoSQL and Hadoop environments than any other vendor. 

How it works

Gazzang zNcrypt™ is a "virtual encrypted filesystem" that shims in at the Linux kernel and is transparent to the database and applications that sit above the filesystem. Data is encrypted "on the fly" as it's written to disk and decrypted when called back by the application. The solution leverages process-based access controls (ACLs) that ensure only authorized, trusted processes can access the data. By restricting data access to certain processes rather than users or roles, you can prevent super users like root from accessing data they don't necessarily need to see. 

Gazzang zTrustee™ is a software-based key manager that secures and manages the keys separate from the encrypted data. This helps ensure a data breach doesn't also result in the loss of the encryption key. Remember, encryption is only as strong as the security of the encryption key. A compromised or weak key is all that's necessary for an unauthorized user or hacker to decrypt and access your sensitive data.

The Gazzang key manager allows the data owner to wrap several layers of policy around the key to prevent unauthorized access. For example, you can set limits on how many times a key can be retrieved or set a specific window of time at which the key might be available.  A unique function of zTrustee is the ability to allow people to authorize or deny key retrievals. These individuals can only determine whether a key should be released, but never actually see the key. You can learn much more about zTrustee by visiting:

Why encrypt Informix?

Earlier in this blog, I mentioned at-rest encryption is mandatory for meeting certain compliance requirements. But if you're using Informix to manage data on behalf of your end customers, chances are they're expecting you to encrypt everywhere and anywhere as well. We work with a number of companies that tell us they could neither have won new business had they not encrypted customer data.

Let us show you how we can help encrypt your Informix data, whether it's in a public, private or hybrid cloud or on premise. Shoot us an email at or register for a complimentary demo and trial. 

Monday, 24 February 2014 16:42

South by Security

Written by

SXSW session picks for the security-minded digital professional (plus, guarding your PII & PCI at the show)

SXSW Interactive is chock full of folks the Wall Street Journal calls “some of the most social, tech-savvy and innovative early adopters you’ll meet.” The best of these digital professionals are also security-&-privacy minded, and we aim to keep these issues at the forefront with our curated list of the best data security sessions at “South By” - including session-specific hashtags for everyone participating from back home. Although, if you don’t attend, how will you visit the Bitcoin ATM?

First, though a couple of tips regarding PII & PCI for attendees:

Personally-identifiable information (PII)

Free food, free beer, free t-shirts, and more can all be had at SX in exchange for just tweeting, checking in, sending a text, posting a selfie, etc. Because - of course - attendees’ personally-identifiable information (PII) is a coveted “product” of the show. For an in-depth discussion of this phenomenon and its repercussions, attend the session “Dear Taco Vendor, How Are You Securing My Data?” with Gazzang Director David Tishgart. As part of the “Core Conversation > Social and Privacy” track on Saturday, March 8th, he’ll cover what data is being collected, where it’s being stored, who’s buying, and who’s selling - and who is accountable for the whereabouts of your PII. #Data4Food.

Payment card information (PCI)

Only 11% of companies actually comply with all aspects of the PCI-DSS standards and SX startups probably represent more the Wild West than average. Before you consider handing over your card details (or your API keys) to that lovely new friend you met at SX, refresh your memory regarding the 12-step program that makes for strong payment data security, for your systems and theirs. Lucky for you, we just collaborated with DataStax (of Cassandra database fame) on a walk-through regarding achieving compliance with your new app: Enabling PCI-DSS Compliance on DataStax Enterprise.


20+ SXSW session picks for the digital security pro

Whether you are headed to Austin next week or just want to get a glimpse of what’s in store, here are our top picks for data security pros. We’ve included session-specific Twitter hashtags so you can live vicariously through attendees’ tweets, even if you can’t make it to Austin.

Friday, March 7th

Saturday, March 8th

Sunday, March 9th

Monday, March 10th

Tuesday, March 11th


Tweet at us and let us here at Gazzang know which sessions you have on your "can't miss" list for this year's South By Southwest.

The who's who of Big Data were out in full force at Strata last week, and like the fall Strata/Hadoop World event in New York, the Santa Clara showcase did not disappoint. 

Allow me to share a few thoughts, straight from the Gazzang booth, which occupied prime real estate right by the food and beverage area: strata og200x200

1) The hype around big data has died down… a lot. Svetlana Sicular of Gartner famously (or infamously) noted early last year, big data is descending into the "Trough of Disillusionment." While that sounds awful on the surface, it's actually a sign of a maturing market. It means all the talk and chest-beating about big data is waning, and the actual tools and technologies associated with the space are starting to yield results. 

The sessions at this year's conference bore that out.  In past events, sessions were dominated by "how-to's" on the latest big data platforms and applications. This year, we heard more from customers and consumers of big data. Sure, you expect to see organizations like Comcast, Netflix and Twitter at Strata, but how about the inventor of ollie, a popular skateboarding trick that I nearly killed myself trying to pull off in the mid 80s?

One session that particularly stood out was GE's talk on the Industrial Internet. Want a use case for big data, and Hadoop in particular? Look no further than how GE is enabling industrial devices (turbines, jet engines, locomotives) to connect and report back on their health, so no machine ever has to be taken offline. It's amazing to think about where this might lead.  

2) The shift from big data hype to production is good for Gazzang as well. While we love to engage with organizations as early as possible in the big data buying process, the fact is that most companies don't think about data security until they start to work with sensitive, production-stage data. In years past, we'd get asked questions about whether we integrate with Hadoop, Cassandra, Mongo, Couch and Riak (by the way, the answer is yes, we do). This year, we heard from dozens of attendees about in-flight big data projects that require at-rest security. Quick shout out to our partners, Rackspace, Cloudera, Hortonworks, Pivotal, DataStax, MongoDB, sqrrl, IBM, Amazon, Basho, Couchbase and Intel for sending them our way. 

3) Speaking of partners, Gazzang made and participated in a number of announcements related to our work with Big Data and Cloud leaders:

DataStax Announces Partner Network Program

Nebula Delivers Multisite Cassandra Workloads Across Private and Public Clouds

Gazzang Provides Data Encryption for Pivotal HD

Our goal is to provide customers with the most comprehensive and proven data security solutions no matter what big data platform(s) they choose. I believe our depth and breadth of experience in these environments is critical to Gazzang being recognized as "The Big Data Security Experts."   

8386650704) What good is a trade show without any fun? For the entire run of the show, the Love Potion Amphora Art/Music Bus (yes, a real thing) was parked right behind our booth. Imagine trying to hold a deep conversation about filesystem encryption with THIS over your shoulder. Despite the distraction, we did manage to make some waves of our own, and we even walked away with an award, courtesy of our friends at Forbes. 

Looking forward to seeing you all back in New York later this year.

Wednesday, 12 February 2014 00:00

Remove the Risk: Share Key Management Responsibility

Written by

What if I told you there’s an easy way to drastically decrease your security risk? No tricks involved.

If you are a SaaS provider, you can share encryption key management responsibility with your customer - the data owner - and give them the ability to revoke the key at anytime. Sounds simple enough, right? Here's why this works.

Offering encryption as a value-added service on top of your application makes a strong statement about the importance you place on protecting customer data. It can be what differentiates an enterprise application from a standard one, and it's likely a requiement if you ever want to work with banks or healthcare organzations.

However, if you also have full control of the encryption keys, that puts an unnecessary management burden on your security team and leaves customer data at risk of being accessed by an unauthorized party. The organization that manages and sets the encryption key policies can ultimately deem who and what can access the data. So, keeping customer privacy and confidentiality in mind, consider whether your employees need to access customer data outside of the few circumstances where it's necessary to perform their job. What happens when a key is misplaced or falls into the hands of a rogue employee? What happens if a subpoena is issued for the encrypted data? How can you be certain your key service meets customer SLAs and compliance? 

Do you want that liability? Do your customers want you to have that liability?

One way around all this is to put your customer's encryption keys in an escrow service. Together you can determine what data the SaaS application needs access to and set policies accordingly. If there's concern over data breach, unauthorized access or the customer simply wishes to discontinue the service, they have the option to shred or revoke the key, rendering the data benign and completely useless. 

We provide this service to a variety of SaaS companies today, and we can do it for you too. 


Ross McDonald, Gazzang Implementation Engineer

In my role as implementation engineer at Gazzang, I'm often asked to install and configure our data security solutions on Cassandra and DataStax Enterprise environments. In fact, those database systems are among the fastest growing in our customer base. 

Protecting sensitive data in NoSQL is critical. The good news is it's not that difficult. We always recommend encrypting data at rest and practicing good key management, but here are a few other security best practices you should consider: 

  • Deploy active protection from malicious system root/administrative user access to sensitive data on your deployments. This ensures that even in the event of a compromised administrative user, your sensitive data is not at risk.

As a best practice, make sure that the Cassandra/Datastax Enterprise process on each node has been isolated by running under its own user(s) and group(s).

  • Did you know sensitive information can leak through completely innocent logs and applications? That's why it's important to secure the logs on the node system as well as any potential log-aggregators in your network (Ganglia, Splunk, Nagios, Zabbix, etc), that could pick up sensitive information. 

System monitoring tools are great for monitoring the health of your cluster, but be aware to take the necessary precautions to secure the monitoring client in such a way that it can't cause trouble. Even if you think your cluster is locked down, check it again.

  • As cloud adoption increases, so does the amount of unprotected backups that are being spun up in different environments. Ensuring your backups are complete and secured will lead to less downtime when you can least afford it.

Take backups of the Cassandra SSTable files regularly, and make sure to take the necessary precautions in order to secure them. It only takes one backup to fall into the wrong hands to cause damage.

I hope these tips help you stand up a more secure Cassandra environment. For any other questions, please don't hesitate to contact us at

Page 1 of 18