The Internet of Things (IoT) has triggered the edge computing. The data generated by massive terminal devices is collected, de-duplicated and processed in the edge data center, which cuts down the amount of data uploaded and effectively alleviates the storage and bandwidth pressure on cloud data centers. Moreover, as the computing nodes are closer to the data sources or users, edge computing can effectively lower the latency of data processing and improve the real-time performance of businesses.
Edge data centers require a 24/7 uninterruptible system that can build high throughput capacity in a limited space. The storage system deployed in the edge data center, on one hand, needs to complete parallel write of massive terminal data, and on the other hand, it is subject to frequent reads by the system for data processing. Therefore, strict requirements are put forward for the throughput and reliability of hard drives.
There’s a customer who has an innovative one-stop edge computing product—ECM, and provides edge computing services with low latency, high availability and low cost. In terms of the hardware, with Mini T-block as the carrier and PBlaze Series NVMe SSD as the main storage device, computing acceleration is realized backed with the enterprise-class CPU.
PBlaze Series NVMe SSD presents extremely low read and write latencies of 72μs and 11μs respectively and offers up to 1,100K IOPS. It provides data prefetching and caching services to help edge data centers overcome problems such as high latency and network dependence arising from long-distance data transmission of cloud storage.
By replacing six SATA SSDs in the original server with a NVMe SSD, both the capacity and performance are improved, which facilitates flexible deployment, and guarantees real-time data storage and access for critical tasks of edge computing.
The deployment of PBlaze Series NVMe SSD substantially reduces the volume of data switching between the edge node services and the hub, which cuts down the cost of long-distance network transmission while improving the performance.
In order to meet the strict requirements of the CDN server for response latency, the server using AMD EPYC platform + Memblaze’s NVMe SSD presents obvious performance advantages, thus greatly lowering the latency of users’ data access and improving the access efficiency of hotspot data.
Relying on outstanding I/O performance, the advantageous multicore of AMD EPYC is better leveraged, which lowers customers’ CAPEX (capital expenditure) and future OPEX (operating expense).
The AMD EPYC platform, with up to 128 PCIe Lanes, can connect more devices such as the NVMe SSD, GPU, and high-speed Ethernet card. Components like PCIe Switch are reduced, which simplifies the system architecture and brings higher performance and stability.
EPYC adopts an embedded security processor to provide a range of functions such as secure startup, memory encryption, and secure virtualization. Moreover, Memblaze’s NVMe SSD guarantees data accuracy and security with end-to-end data protection and AES data encryption and other technologies.
