Key Takeaways
- CTERA designs its platform around three needs that don’t change as technology evolves: the need for data security, the need for fast data access, and the need to access different types of data across the enterprise.
- Security requirements evolve to meet demands but the need for security does not. For example, FIPS 140-2 certificates are no longer active after September 21, 2026. CTERA’s architecture is built so new cryptographic and compliance standards can be incorporated and deployed enterprise-wide, as they become the standard.
- Data in the cloud will never be as fast to access as data held locally. CTERA Edge Filers cache frequently accessed data locally to close that gap.
The Three Constants Behind CTERA's Architecture
With the rise of AI, the world is changing faster than ever. What once took years now takes months or weeks, and what took months or weeks can now be accomplished in days. In this environment, who can say what the next year will bring, let alone the next decade.
It brings to mind a question that Jeff Bezos likes to pose.
Though he’s often asked what will change in the next 10 years, he prefers to think about what won’t change. And based on this perspective, he and his team have built AWS’s long-term strategy around three cornerstones that will remain constant no matter what else happens: faster, cheaper, and a vast selection. No customer will ever ask for less of any of them.
CTERA is not AWS, and our strategic requirements are different. But we can apply the same thinking. What won’t change for CTERA?
- The need for data security.
- The need for fast access to data.
- The need to access different types of data.
By focusing on these immutable requirements, CTERA has built a platform that can adapt to an ever-changing future.
How These Constants Shape CTERA Strategy
CTERA’s strategy rests on the three constants that will hold true regardless of what comes next: strong security, fast access to data, and unified access across data types.
A Security Architecture Built for the Unknown
Let’s address security first. As time has progressed, security requirements have changed to keep pace not just with existing threats but emerging ones as well. When a new security standard is introduced, the CTERA infrastructure enables it to be covered, whether it be FIPS 140-3 compliance (which CTERA supports), replacing the older FIPS 140-2 standard, or FedRAMP compliance, or whatever else comes down the line. The underlying security architecture used by CTERA products means that additional capabilities can be easily incorporated.
CTERA security is not just an add-on but a built-in requirement. Security doesn’t cover just compliance and governance; it’s also about protecting the data. For example, we know that the ability to hack systems using AI will only increase as time marches on. CTERA has built its ransomware protection bearing this in mind. Using honeypot exfiltration and behavioral and machine learning analysis rather than signature-based detection means that as attacks become more sophisticated, so does our product’s ability to prevent them.
Fast Access to Data, Anywhere, Anytime
Second, fast access to data. Cloud storage is often an answer to cheaper and more reliable storage, one of Jeff Bezos’s three unchanging principles. But no matter how fast AWS or any other cloud provider makes cloud compute or storage, networking physics means accessing data in the cloud from on-premises locations won’t be as fast as local access. CTERA recognized this from day one, so alongside security, building fast access via edge appliances was a no-brainer. Caching frequently accessed data locally from all the data stored in the cloud makes access to that data fast.
Unified Access to Different Types of Data
Third, access to different types of data. Data has been around for thousands, if not hundreds of thousands, of years, from Stone Age carvings and papyrus scrolls to books to the digital age in which we now find ourselves. The ability to process this data has changed, but the need for it hasn’t.
Accessing unstructured data stored in silos of object buckets or blobs or file systems requires a different mechanism than accessing structured data. While humans use traditional file storage to collaborate on documents, AI pipelines require high-scale object storage to train models. File systems support collaboration and applications, while object storage commonly powers analytics and AI workloads, such as data lake and lakehouse architectures. Connecting the two typically requires additional complexity, like gateways, data duplication, or complex workflows that increase cost and latency.
By recognizing this and creating a patented object-to-file mapping (CTERA Fusion Direct) to unify file and object storage, CTERA has brought the future closer. As AI workloads continue to evolve, CTERA’s unified approach ensures file and object data will remain accessible together, both now and in the future.
Building for the Unknown
Do we know what the future holds?
It used to be easier to predict. But with architecture designed around the things we know will always matter, CTERA can adapt to whatever the future holds while continuing to provide secure, fast access to your data, whatever the format.
Frequently Asked Questions
- What is the difference between FIPS 140-2 and FIPS 140-3?
FIPS 140-3 is the current NIST standard for validating cryptographic modules; FIPS 140-2, issued May 25, 2001, is its predecessor. The main difference is alignment — FIPS 140-3 adopts the international ISO/IEC 19790:2012 and ISO/IEC 24759:2017 requirements and testing framework, while 140-2 was a standalone NIST standard.
- Why is cloud data slower to access than an on-premises location?
Distance and network latency. A read from local flash completes in well under a millisecond, while a round trip to a distant cloud region takes tens of milliseconds, and no amount of cloud performance changes how fast light moves through fiber. File protocols compound it: opening and reading a file is a sequence of operations, and over a WAN each file operation adds time. Caching frequently accessed data locally is the standard answer. CTERA stores the full dataset durably in object storage while the active working data stays near the user.
- Do AI workloads need object storage or file storage?
Both, usually at different stages of the same workflow. Large-scale training and analytics often read from object storage over S3, while applications, AI agents, and the people who create the data, commonly work in file semantics over SMB and NFS. This has led to duplicate copies of data, increasing cost and complexity in AI workflows. Unified file and object architectures such as CTERA Fusion Direct enables the same data to be reached both through both object and file protocols without a copy.
