On the application side, the quest for zero-latency storage is real. In this era where technology is ubiquitous, the multitudinous latency-sensitive applications that surround us require fast and efficient processing of data at massive scale. Providing near-zero latency at such scale is the remaining storage challenge and by extension, the most pressing technology challenge for web-scale data centers.
Scalability
-
Scale linearly to any size, data center-readiness
-
Client-side intelligence means data services scale as you add clients
-
Locking is 100% distributed across targets and clients and scales as you add nodes
-
No centralized metadata manager means no scalability bottleneck
New-generation flash media, such as NVMe and SCM, are moving the bar on storage latency. Single-digit μs (microseconds) latency is a reality when used locally. This is setting expectations for application developers, who now get much better performance from one local NVMe flash device than an entire enterprise-grade all-flash array.
But the real challenge is to share NVMe across the network, deploy NVMe at scale with the same low-latency as when used locally. This cannot be done with traditional controller-based architectures as those can only do low levels of IO processing before their bottleneck-design slows down, increases latency and eventually tops out.
New storage and networking technologies enable scale-out in-server flash architectures.
By deploying NVMe as in-server flash in distributed, shared-nothing architectures, IO processing does not suffer traditional controller limitations: as you scale devices and network bandwidth, you linearly scale performance. Applications benefit from the low-latency of NVMe in a scale-out fashion, leveraging standard networking or new-generation networking such as RDMA for ultra-low latency.
Excelero’s mission is to radically change the way digital businesses design storage infrastructures for low-latency, scale-out applications. Our goal is to enable storage managers to provide shared low-latency storage with efficiencies akin to those of the most efficient data centers.
Flexibility
-
Mix different storage media types, server types and deployment models
-
Use native NVMesh and/or NVMf transports to future-proof your environment
-
Avoid one-type fits all logical volume simplifications
-
Scale Storage and CPU independently
Efficiency
-
Enable high NVMe SSD utilization rates
-
Use standard hardware
-
No CPU Overhead
-
No forklift upgrades
-
Low management cost
Ease of Use
-
Easy to manage & monitor
-
Integration with orchestration layers for provisioning of logical volumes
-
Storage access API – block is the common building block
Performance
-
Leverage the full performance of your NVMe SSD at scale, over the network
-
True convergence to ensure deterministic application performance – no noisy neighbors
-
Leverage high IOPS, high bandwidth or mixed