Watch on YouTube
Watch on Vimeo
After an introduction by CEO and Co-Founder Lior Gal, Yaniv Romem, CTO and Co-Founder, lays out Excelero’s vision: to enable businesses to build an efficient, high performance, scale-out storage infrastructure. They built their solution predicting that applications increasingly expect NVMe capabilities from storage, while enterprises increasingly deploy 25/50/100G and with it RDMA. Their product, NVMesh just launched. It’s a software layer wherein NVMe devices are pooled and logical volumes are created on top of that.
Personnel: Lior Gal, Yaniv Romem
Watch on YouTube
Watch on Vimeo
Josh Goldenhar, VP Products, goes over some of the use cases for Excelero’s NVMesh. This solution is unique in that it doesn’t use any CPU on the target device, and is capable of millions of IOPS. Josh outlines how this can be used to bring hardware homogeneity to the cloud by standardizing on hardware, and giving the flexibility to actually move storage to compute. This is ideally served for applications from real-time analytics, to databases, and container orchestration.
Personnel: Josh Goldenhar
Watch on YouTube
Watch on Vimeo
Yaniv Romem, CTO & Co-Founder, reviews the architecture and software component of their NVMesh solution. Their solution uses the client CPU to avoid hindering application performance inherent in using the target CPU. Instead, the target RDMA NIC writes directly into the NVMe drive, making the data path effectively direct from the application to the target drive. This is all managed by an independent control layer. All of this is done with standard RDMA NICs running on Linux.
Personnel: Yaniv Romem
Watch on YouTube
Watch on Vimeo
Josh Goldenhar, VP Products, demonstrates their NVMesh solution in action. They’ve tested this with the major NVMe vendors, and allows you to mix and match drives as you see fit. The overall solution adds a mere 5 microseconds to reads and writes. Using relatively low-end Intel Xeon processors, Josh demonstrates getting remote IO response time of 94 microseconds vs 88 microseconds locally, and 4.5 million random read 4k IOPS, on a server with $13,000 worth of hardware.
Personnel: Josh Goldenhar
Watch on YouTube
Watch on Vimeo
Yuval Bachar, Principal Engineer of Architecture and Strategy at LinkedIn, discusses the Open19 project started by the company to create an that can fit server, storage, and networking into a 19″ rack, that is applicable to environments and organizations of all sizes. He reviews the hardware building block principals of this standard. Excelero is involved and supporting the solution.
Personnel: Yuval Bachar
Thank you for being part of the Tech Field Day community! Our mailing list is a great way to stay up to date on our events and technical content, and we appreciate your signup.
We promise that we’ll never spam you, send ads, or sell your information. This list will only be used to communicate with our community about our events and content. And we’ll limit it to no more than one message per week.
Although we only need your email address, it would be nice if you provided a little more information to help us get to know you better!
