The explosive growth of data requires enterprises to scale up their IT infrastructure to meet the demands set by IoT, big data, and artificial intelligence platforms. In an exclusive interaction with Analytics Insight, Lior Gal, CEO and Co-Founder of Excelero shares how the company’s data storage platform used with AI, machine learning and analytics workloads enables users to iterate endlessly on hundreds of terabytes of data.
Analytics Insight: Kindly brief us about Excelero, its specialization and the services it offers.
Lior: Excelero enables enterprises and service providers to design scale-out storage infrastructures leveraging standard servers and high-performance flash storage. Founded in 2014 by a team of storage veterans and inspired by the tech giants’ shared-nothing architectures for web-scale applications, the company has designed a software-defined block storage solution that meets performance and scalability requirements of the largest web-scale and enterprise applications.
With Excelero’s NVMesh, customers can build distributed, high-performance server SAN for mixed application workloads. Customers benefit from the performance of local flash, with the convenience of centralized storage while avoiding proprietary hardware lock-in and reducing the overall storage TCO. The solution has been deployed for hyper-scale Industrial IoT services, machine learning applications and massive-scale simulation visualization.
Analytics Insight: With what mission Excelero was set up? In short, tell us about your journey since inception of the company?
Lior: Excelero was inspired by how Amazon, Facebook, Google, LinkedIn and other tech giants have redefined IT for web-scale applications. These companies don’t buy proprietary hardware and software to run their operations. They leverage standard servers and “shared-nothing” architectures to ensure maximum operational efficiency and flexibility, and to provide the highest reliability.
Enterprises and service providers want to do the same. For storage, this means they want to deploy unified scale-out storage infrastructures leveraging standard servers, also described as Software-Defined Storage (SDS). Customers also want to consolidate their current storage silos into unified, scale-out, flexible block storage pools on standard servers, known as Server SAN. Consolidating storage tiers avoids the need to overprovision storage, do forklift upgrades and lowers the overall TCO.
NVMe flash has great potential for SDS deployments, delivering better performance and reduced latency, but there is a caveat: the storage needs to be used locally. Applications are limited by the capacity and performance of the NVMe drives within each server. It’s virtually impossible to level out utilization across your entire infrastructure, which results in NVMe capacity and performance waste.
Excelero’s Software-Defined Storage platform NVMesh® enables customers to design Server SAN infrastructures for the most demanding enterprise and cloud-scale applications, leveraging standard servers and multiple tiers of flash. NVMesh enables customers to pool flash over a network at local speeds and deliver a true converged infrastructure by logically disaggregating storage from compute. It bypasses the CPU and avoids noisy neighbors, which is ideal for scale-out applications. This approach provides deterministic performance for applications and enables customers to maximize the utilization of their flash drives. Excelero’s NVMesh supports both converged and disaggregated architectures.
Analytics Insight: Brief us about your role and contributions towards the company and the industry.
Lior: I am a seasoned enterprise IT leader who has grown numerous storage innovators from concept to proven product and multi-million dollar revenue streams and, in several cases, successful exits. Beginning in technical roles at EMC and Amdocs, I led sales at Exanet, Illuminator, Crescendo Networks, and most recently was global vice president at Data Direct Networks, where I achieved dramatic revenue growth from the world’s largest, high-performance web and cloud customers.
Analytics Insight: Tell us how Excelero is contributing to the Big Data Analytics industry and how the company is benefiting the clients.
Lior: Real-time analytics workloads demand the lowest levels of latency and high I/OPS storage at scale. The same holds for IoT, Artificial Intelligence (AI), Machine Learning (ML), and cognitive computing deployments – and applications where database volumes in the data repositories or within Database as a Service (DBaaS) deployments. Traditional data center solutions aren’t built to scale or perform at optimal efficiency with these use cases.
As a result, companies with any form of Big Data applications are rethinking their storage strategies. By keeping data closer to where it is needed, applications spent less time is spent waiting for I/Os from shared storage. That’s the heart of the Excelero approach. Excelero’s NVMesh software eliminates I/O wait‐time and associated server CPU overhead by reducing the number of expensive network hops to shared fast pooled storage. Applications benefit from NVMesh with improved relative performance and lower latency.
The benefits are particularly significant with the database. With NVMesh, extracts, clones, and large full table scans run quicker, database tables, indices, and other entities create as well as load faster. Instead of spending time creating and managing specific indices, materialized or different views, administrators can apply those resources and time to more productive business functions. This means data centers need less server and storage hardware and can readily boost performance.
Analytics Insight: Kindly mention some of the major challenges Excelero has faced till now.
Lior: NVMe has seen very fast adoption with small deployments involving just a few volumes. But surprisingly as recently as 2016 many large enterprises had a hard time buying the NVMe Flash storage that was suitable across thousands of servers at once. Projects would often be delayed due to availability. Fortunately, that issue is now ameliorated as next-generation NVMe storage options emerge and results detail that they are indeed solid and providing demonstrable ROI.
Then, of course, there’s the typical startup challenge that we at Excelero face as does everyone – the need to establish trust as a critical first step to early adoption. It’s especially true in our case since we store data for business-critical applications. To that end, Excelero made four strategic moves that, looking back, we believe were a fundamental on-ramp to our strong momentum.
1. We launched almost 3 years after we were founded, giving us a handful of major name, paying customers – and compiling a set of early adopters including GE Predix, NASA Ames, and Hulu right out of the gate.
2. We emphasized innovation, quickly establishing 15 patented or patent-pending technologies from a team of stellar technologists so we established defensible IP.
3. We skew our marketing and sales approach toward strategic partners who can deliver volume and make us part of a total solution versus toward a direct sales emphasis to the enterprise.
4. We raised funds ahead of when they were needed when market conditions were extremely favorable -giving us a cushion where we can continue to refine the solution and add features and data services that major players have come to expect.
Analytics Insight: Mention some of the awards, achievements, recognitions that you feel are notable and valuable for the company.
Lior: Excelero has commandeered an impressive lineup of industry awards ever since our stealth debut in March 2017. These include being named the Gold Medal winner as the Software-Defined Storage Product of the Year in the 2017 Products of the Year awards from Storage Magazine and SearchStorage; Best in Show award as Most Innovative Flash Memory at the Flash Memory Summit, August 2017; ‘Storage Innovation of the Year’ at the SVC Awards, November 2017; and nationally ranked on the Entrepreneur360™ List of the best privately held entrepreneurial companies in the US, as compiled by Entrepreneur magazine. We’re extremely proud that our very competitive industry recognizes the innovative approach we have taken to shared NVMe storage at local performance.
Analytics Insight: What does the Big Data Analytics market need?
Lior: Top-performing companies are more effective in every aspect of data analytics, and typically they have optimized infrastructures and applications to capture, store and analyze more data faster to make better business decisions. Applications like Splunk, SAS and SAP Hana enable customers to gain impressive insights in very large datasets. But these data sets are growing and customers are hungry for more. They need to analyze more data, faster. The financial impact of being able to predict events well before they happen can be huge.
Next-generation analytics essentially requires two fundamental changes: faster storage, and more scalable architectures. The latest generation of NVMe flash devices already solves the speed part of the problem. The remaining issue, however, is that this performance can only be consumed locally – and so NVMe is not a great fit for traditional array-based architectures.
Excelero’s NVMesh addresses this second issue simply, and elegantly, by enabling customers to deploy NVMe at massive scale, without adding latency or data locality. That’s why we’re achieving such tremendous momentum in analytics applications including Industrial IoT, machine learning, and user behavior analytics applications. Customers obtain the performance of local flash, with the convenience of centralized storage and the cost savings of standard hardware. Analytics applications can access the block volumes directly or through a file system, which makes integration very straightforward.
Analytics Insight: How are disruptive innovations impacting this sector?
Lior: Everyone cites AI, machine learning and Big Data as key disruptors in the data center, without looking deeper to what powers them. Running AI on yesteryear’s networking and storage simply cannot let it deliver – period. Next-gen workloads demand next-gen networking and storage approaches. As result companies are overhauling their data centers for the scale-out generation. They’re replacing older networking standards with 25/50/100Gb Ethernet, moving to NVMe SSDs, using commodity hardware not custom-purposed servers and storage, and looking at sharing unused NVMe instead of continually buying more hardware and software as needs grow.
Analytics Insight: How does Excelero’s expertise uncover patterns with AI and ML?
Lior: Excelero’s approach makes it faster, easier and more cost-effective for researchers to access mission-critical AI and machine learning applications – without their organizations needing the level of internal expertise to support them.
AI, machine language, cognitive computing and similar applications generate, process and store massive amounts of data every day. Their intensive data volumes are I/O performance drivers, including both small and large transactions or other I/O operations. When there’s a server or storage I/O bottleneck, end-user performance suffers, and the vital work of image analysis, medical research, IIoT monitoring, eCommerce, entertainment content production and delivery, and other uses can be slowed or interrupted.
Excelero’s NVMesh used with AI, machine learning and analytics workloads enables users to iterate endlessly on hundreds of terabytes of data – expediting both time to results and time to market with their ideas. Organizations avoid the need for database administrators (DBAs), significant data center equipment and conversion teams to write scripts for performing analytics queries. In some instances Excelero cut deployment time by 80% and delivered 3x the performance of previous scale-out storage solutions – for a fraction of the price.
Analytics Insight: Where are we going from here?
Lior: An estimated 70 – 80% of enterprises say they intend to adopt a web-scale IT architectural approach by 2027, according to Intel data – enterprises for whom Excelero’s 100% software-only server SAN solution is a perfect fit.
Excelero’s CEO founded the company after his biggest customers at his last position said the same thing. They made it clear they weren’t buying more hardware/software boxes – from him or anyone. They wanted off the shelf, commodity hardware — and planned to spend as little as possible on custom hardware and simplify their operations while putting the emphasis on software as their key differentiation, and making sure whatever they deployed was highly scaleable, agile and multi-functional. Excelero is all about fulfilling that need. The market’s rapid endorsement of our approach has provided tremendous momentum for us in 2018.