Brytlyt’s ground-breaking serverless GPU analytics platform is a game-changer in the analytics market. The usage-based model is the most cost-efficient way to run analytics solutions. This first-of-its-kind model for GPU can save businesses up to 90% of their hardware costs – breaking down one of the most significant and long-standing barriers to GPU-accelerated analytics.
Finally making GPU technology accessible
Brytlyt’s usage model is completely unique compared to typical GPU solutions. Using the same usage-based pricing pioneered by popular cloud-based providers like Snowflake, Brytlyt’s serverless model provides end-to-end access to accelerated analytic capabilities from a simple browser log in.
The user experience feels just like a typical GPU solution, completely decoupled from what happens in the background with the serverless commissioning and decommissioning of hardware. A user’s session state is always ready and waiting for their log in, hardware is decommissioned when the user logs out and instantly recommissioned when they log in. Hardware is only ever allocated for a specific user workload, when it’s needed.
This generates huge cost-savings compared to the status quo, where hardware is running 24/7 regardless of how many users are logged in. A serverless solution is specifically designed to take advantage of the variable nature of analytics workloads. When Brytlyt analysed real-world usage profiles of their GPU solutions over time, they found there were short peaks of intense workload, followed by long periods of little or no usage. On average an individual user only accesses the platform 10% in a 24-hour period.
Typically, hardware for a solution is sized according to the peak workload, resulting in an investment which is often massively underutilised most of the time. A serverless solution is very different in that it gives users access to the hardware they need as they need it. Hardware is only paid for when it is used, resulting in saving that often approach 90%. In addition to the hardware saving, a serverless solution also massively reduces maintenance overheads when compared to dedicated or pay-as-you-go hardware.
Recommended AI: Microsoft 365 Security Features Protect Business Data from Evolving Threats
Pay-as-you-go and serverless are NOT the same thing
Brytlyt’s serverless model is the next step from cloud analytics services. Just like renting a car is very different to catching a taxi, a pay-as-you-go model is very different to a serverless model.
In a serverless model, not only does the software provide the analytic tools, it also dynamically manages the allocation and de-allocation of the hardware as needed without user intervention. In a pay-as-you-go model, a user must actively manage the hardware provision and teardown. This additional overhead massively limits the cost benefits of pay-as-you-go solutions.
Buying hardware outright is equivalent to owning a car. One gets something big enough for the entire family, accommodating their highest capacity journeys, even though the most common journey is most likely a solo commute. And the car sits on the driveway most of the time anyway!
Like car rental, cloud solutions are more flexible, but not completely cost-efficient. While rental models allow flexible, short-term usage, there is no saving when the rental car is parked outside – in this analogy the cloud hardware.
Recommended AI: Consider Your DOOH Buying Methods Wisely: Direct Sales vs. Programmatic Buying
A serverless (or taxi) model is completely aligned with actual usage. Users only pay for the exact duration or distance of their trip. Why pay for hardware that they’re not using?
The only difference in this analogy is that serverless users don’t have to wait 10 minutes for their taxi to arrive – serverless solutions are available immediately. Unburdened by the capital expense of purchasing GPU hardware, businesses of all sizes can now access and flexibly scale their GPU technology. Users are free to adopt more advanced analytics techniques, from exploratory and predictive analyses to performing rapid billion-row workloads and ad hoc queries.
Brytlyt’s serverless is already gaining momentum
What truly sets Brytlyt’s platform apart from the other cloud solutions is the impressive technological challenge the company has had to overcome to make this possible for large GPU analytics applications. While other companies are only starting to figure out complex workloads on GPU, Brytlyt is years ahead, making complex GPU-powered analytics accessible and highly cost-efficient within the cloud.
Already proving to be extremely popular, with industry leading clients such as Accenture, Brytlyt are releasing a new version of their serverless GPU platform in November following several other exciting developments, including an update to their visualisation tool. This is an exciting time for the global company who are transforming the landscape for analytics and making deep learning accessible to all.
“The serverless model for GPU really is a ground-breaking evolution for the analytics market. Users get the benefits of our data visualisation and deep learning platform on-demand, without the overhead and complexity of managing their own software and hardware resources. This unlocks a lot of exciting value by transforming how businesses gain insights from their data.” – Richard Heyns, Founder and CEO of Brytlyt
Recommended AI: Understanding the Role of AI in Gaming
[To share your insights with us, please write to sghosh@martechseries.com]
Comments are closed.