This is part one of a five-part blog series that will take a closer look at our new suite of data services that will help our customers to radically simplify file data management at scale. We begin here with an overview of our new NVMe Cached Performance. Future blogs in this series will go more deeply into the other new data services included in this announcement.
Let’s take a look at the NVMe cached performance enhancement as part of this week’s announcement.
Qumulo now delivers NVMe performance for the price of disk.
The benefits of NVMe have historically been accessible only to specialized workloads and to organizations with large budgets. There are many workloads that can benefit from the low latency and fast streaming performance of NVMe but require more storage capacity than can be fit in their budget to leave the data on NVMe permanently.
We believe that managing your unstructured file data should be easy. Tiering is not the solution
Many organizations have tried to solve the need for low latency, fast streaming coupled with high levels of capacity by creating tiers of storage: a high performance tier, a warm storage tier and a cold storage tier. Managing these tiers, which are often manual or policy based, turned out to be too complex and data was often not easy to access when needed.
Tiering can be complex in 2 key ways:
- The people working with the data (researcher, artists, data scientists) depend solely on the correct policy for their data being set and managed by the storage administors. If the policy scheduled jobs run on schedule and complete successfully, everything is fine, but when they aren’t, access is slow or applications time out. Any changes to these policies require working with IT to correct them, and once they are corrected, it can take a long time for the system to make the adjustments.
- The core responsibility is now on the storage administrators, who often are busy managing other parts of the infrastructure and don’t have the time to micro-manage these policies. The storage administrator who has been able to purchase a flash tier has to walk a fine line of filling up the performance system as full as safely possible, to get the most value out of the system. This is a tough job and being wrong is expensive, in both wasted time and impact to those trying to do work with the data on the system.
Qumulo believes there is a better way
There really is a better way to solve this problem. Let the software, the intelligent and automated software, do the work!
We designed the file data platform so that every byte of data written to Qumulo was written to SSD. Qumulo has been shipping systems that use both SSDs for reads and writes + HDDs for over 6 years. This has helped our users get both a real time architecture with data visibility and the performance benefits of SSDs for all of our users.
Machine Learning Cache
Qumulo’s machine learning cache which was originally designed for SSDs has been optimized for NVMe. Just as Qumulo continually enhances and improves the entire file data platform, the machine learning chase is continuously being improved. The machine learning cache now leverages both pattern matching to read ahead data, as well as block level activity for heat analysis. The intelligence actually looks at not only clients, but sessions on that client to determine what caching mechanisms will work best. The machine learning cache enables over 87% of data to be read from memo or NVMe SSDs in a hybrid environment.
This enables users to have the best experience while making the best use of the SSD cache in their system. For more details, check out Tommy Unger’s blog post.
Qumulo Hybrid NVMe
Qumulo software now supports an NVMe SSD + HDD architecture introducing two new systems, the C-192T and the C-432T. These systems significantly lower the cost of performance for customers. By using NVMe in front of the largest HDDs in the market, we provide our customers a better experience and something that fits within tight budgets. All writes land on NVMe SSDs, greatly increasing performance. By using our machine learning-based read caching, where 80+% of data is read from SSD based upon actual fleet data, those reads also get way faster. The net result is a large file system where users have much faster access to their data at one-third the cost of All-NVMe solutions.
What is the benefit to users
Now more people can have NVMe speed and lower latency (30% more responsive) access to their data. The transition to accessing files via NVMe storage is similar to that of moving your laptop from a HDD to an SSD. For those that remember that transition, it was night and day. You went from booting your computer in 10 minutes to 30 seconds. Once users have a taste of that, they never want to go back. Most of us don’t have to work with the large datasets like those that creative and research people do. We are spoiled by rapid access to our data. These people are living in the era of slow data access, which is horrible. We built this system to enable these people to stop waiting on storage and focus on not just results, but the best results. This enables them to focus on their passion and not waiting for systems.
“The Qumulo Core’s software’s new capability to use NVMe as cache enables us to meet our performance and budget needs by providing great economics from very dense hard drives combined with its great caching capabilities on extremely fast NVMe drives,” said Serkan Yalcin, Director of IT, Infrastructure Dev/Ops, Institute for Health Metrics and Evaluation.
“Qumulo’s intelligent caching, without managing policies, has been great and provides even more value for us with NVMe as the caching layer. These features, combined with their great customer success experience, really empower us to focus on our mission of bringing population health data to the world.”
What is the benefit for storage administrators
This makes purchasing, deployment, and management of storage dead simple. The system automatically takes full advantage of all the resources, enabling them to focus on other parts of the business and their customers fast access to data. No more managing policies, unhappy customers, or wasting resources in moving data. Since Qumulo software is a subscription, administrators and users benefit from our continuous improvements to performance and added features, which are included. We are focused on our customer’s success and not pushing new hardware.
Help us change lives
Learn more about this new architecture and help us simplify the data center and empower people to make the next great movie or discover the next breakthrough without getting a coffee as they wait for their storage.
Learn more
To see these new data services in action, sign up for our Q-Connect exclusive virtual event Dec. 8-16!
At Q-Connect, attendees will have the opportunity to connect live with our technical experts and Qumulo customers to see how our file data platform radically simplifies enterprise data management.
If you’re tired of breaking the bank for performance, having difficulties accessing data, being constrained in leveraging technology advancements, and paying for clunky data security and time-consuming upgrades, attend this event to see how Qumulo can help.
Key takeaways:
- Network and hear from your industry peers
- See Qumulo’s rapid release innovation in action with live demos and open Q&A
- Join our CEO and other Qumulo executives for an informal talk shop happy hour
Contact us if you’d like to set up a meeting or request a demo. And subscribe to our blog for more helpful best practices and resources!
