Edgecore 25.6T AI Switch AIS800-32D - Ảnh 2

Edgecore 25.6T AI Switch AIS800-32D - Ảnh 3

Edgecore 25.6T AI Switch AIS800-32D

Liên hệ

32 x 800G QSFP-DD800 switch ports with Tomahawk 5
High-performance, low latency switch for high-performance data centers
Up to 30W power budget per port

Add to wishlist

Mô tả

The Edgecore AIS800-32D is a high-performance, low latency switch for high-performance data centers.

Application Scenarios

Spine Switch

A next-generation, highest-capacity switch for data center spine use case. Breakout options include 2 x 400G, 4 x 200G, and 8 x 100G per port, with a maximum of 160 logical ports. Offers reduced cost and power per bit. Scalable and enables migration to 400G leaf connectivity in data centers

AI/ML Clusters

Standards-based (Ethernet) networking for AI/ML training, leveraging low latency and high-throughput RoCEv2. Reduces Job Completion Time (JCT) using the cognitive routing and congestion management capabilities of the switch. Fully programmable telemetry enables sophisticated on-chip applications for heightened network insight and efficient network management.

High-Performance Computing

The large number of high-capacity Ethernet ports enables server interfaces to transition to higher speeds and denser networks.
Enables the virtualization of compute and storage with VxLAN switching and routing.

Cloud DCI

Supports 400G QSFP-DD ZR/OpenZR+/+6dBm ZR+ and future-proof 800G QSFP-DD800 ZR/ZR+/+6dBm for cloud DCI scenarios.

Features

QSFP-DD800 switch ports, each supporting 1 x 800 GbE (100G PAM4), or via breakout cables 2 x 400G GbE, 4 x 200 GbE, or 8 x 100 GbE.
QSFP-DD800 switch ports also support 1 x 400 GbE (50G PAM4), 1 x 100 GbE (NRZ), and via breakout cables 2 x 200 GbE, 4 x 100 GbE, or 8 x 50 GbE.
Up to 30W power budget per QSFP-DD800 port.
Incorporates Broadcom Tomahawk 5 switch series silico
Highest Radix: Up to 160 logical interfaces on a single chip, low latency
Cognitive/adaptive routing and Dynamic Load Balancing (DLB) and Global Load Balancing (GLB)
Advanced shared buffering.
Programmable in-band telemetry.
Supports end-to-end congestion control.
Power efficient due to a monolithic 5nm die.
Hardware-based link failover for network resiliency and reduced job completion time.
Support for SRv6.
Support for VxLAN RIOT
BMC module with serial-over-LAN support
(Optional) SyncE and PTPv2 support with 1PPS, 10 MHz, and ToD connectors on the front panel
Contains e-fuses to protect transceivers and internal components
1 RU form factor
Supports hot/cold aisles with front-to-back/AFO/port intake airflow SKU and back-to-front/AFI/port exhaust airflow SKU
All ports on front; PSUs and fans accessible from rear
Hot-swappable, load-sharing, redundant 2400 W AC/DC PSUs
7 Hot-swappable fan modules with 6+1 redundant fans
Hardware switch pre-loaded with Open Network Install Environment (ONIE) for automated loading of compatible open source and commercial NOS offerings

Specifications

Switch Ports	32 x QSFP-DD800 800GbE
Switch Silicon	BCM78902 Tomahawk 5
Processor	Intel® Xeon® Quad-Core 2.2 GHz
BMC	AST2620 with OpenBMC secured by AST1060 Root of Trust
SPI Flash	64MB x 2
Memory	32GB DDR4 SO-DIMM with ECC
Storage	240GB m.2 2280 NVMe SSD

TPM	TPM2.0 SPI
Rack	1 RU form factor
Timing and Sync	(Optional) 1PPS Port, 10 MHz port, ToD port, SyncE, IEEE 1588v2 PTP
Fans	7 Hot-swappable fan modules with 6+1 redundant fans
Storage Temperature	-40°C to 70°C
Operating Temperature	0° C ~ 40° C (Front-to-Back)0° C ~ 35° C (Back-to-Front)