SeaMicro SM10000-XE
Contact SeaMicro to learn more about how the SM10000-XE can help reduce TCO in your data center.
Download the PDF: 64-bit SM10000-XE
Hardware Specifications: SM10000-XE
Processor | |
| Total No. of Processors per system | 64 |
| No. of Processors per Compute Card | 1 |
No. of Compute Cards per system
| 64 |
Processor Specification
| Intel E3-1260L |
Advanced Processor Features | Intel x86-64, VT-x, HT |
| Memory Capacity | 8, 16, 24, or 32 GB per socket |
| Memory Type | DDR3 ECC SODIMM |
| 10 GbE Ethernet Uplink | |
Max. No. of Ethernet Interfaces
| 16 |
| Min. No. of Ethernet Interfaces | 8 |
| Ethernet Interface Type | 10/100/1000BaseT |
Max. No. of Ethernet Cards
| 8 |
| No. of 1GB Ethernet Ports per Card | 8 |
| Storage | |
| Max. No. of Physical Disks | 64 |
| Min. No. of Physical Disks | 0 |
Max. No. of Storage Cards
| 8 |
| No. of Physical Disks per Card | 8 |
Type of Physical Disks
| 2.5” |
| Hard Disk Drive Options | |
| Enterprise 7200 RPM SATA | 500 or 1000 GB |
| SATA MLC Solid State Disk | 80, 160 or 300 GB |
| Load Balancer | |
| Layer 4 connections per second | 500,000 |
| Max. No. of concurrent connections | 32,000,000 |
Layer 4 aggregate throughput
| 64 Gbps |
| Maximum number of VIPs | 64 |
Load Balancing Methods
| |
Server Health Checks
| |
| Physical Characteristics | |
| Power Supply (AC) | 3+1 Redundant 240V AC Single Phase |
| Management Console | Dual Redundant |
| Out-of-band Ethernet | 10/100/1000BaseT |
| Cooling | Dual Redundant Fan Tray |
| Air Flow | Front to Rear |
| Dimensions (H x W x D) | 17.5 x 19 x 30” |
| System Software | |
Embedded Management | |
| Management RFC Compliance RFC768 UDP RFC793 TCP RFC854 Telnet RFC3164 Syslog | |
Integrated Terminal Server Telnet/SSH access by TCP port, IP address or server name | |
| General IPv4 Protocols RFC791 IPv4 RFC792 ICMP RFC826 ARP RFC1027 Proxy ARP RFC1035 DNS (Client) RFC1519 CIDR RFC1542 BOOTP (PXE Client and Relay) RFC2131 DHCP (Server and Relay) | |
| SNMP MIB Support RFC1213 MIB RFC1215 TRAP-MIB RFC2863 MIB SNMPv2 MIB SEAMICRO Enterprise MIB SEAMICRO TRAP MIB | |
| Environmental | |
| Operating Temperature | 50° to 95°F (10° to 35°C) |
| Non-operating Temperature | –40° to 149°F (–40° to 65°C) |
| Operating Humidity | 5 to 93% non-condensing |
| Non-operating Humidity | 5 to 93% non-condensing |
| Warranty Information | |
| Hardware | 3 year |
| RoHS | |
| All SeaMicro components are EU RoHS compliant. |
Download the PDF: 64-bit SM10000-XE
Uses 1/2 the power, takes 1/3 the space, and delivers 12x the bandwidth of today's best-in-class servers
The SeaMicro SM10000-XE is the Highest-Density, Most Energy-Efficient Xeon™ Server Ever Built — Uses Intel™ Sandy Bridge–based Xeon® processors
Designed to replace 32 1-rack-unit (RU), dual-socket servers, the SM10000-XE integrates 64 Intel Sandy Bridge–based Xeon processors, storage, top-of-rack switching, server management, and application load balancing into a single 10-RU system.
Optimized for application-tier workloads like Java, PHP, memcached and noSQL databases, the SM10000-XE uses 1/2 the power, takes 1/3 the space, and delivers 12 times the bandwidth of the existing best-in-class server without requiring modifications to existing software.
The SM10000-XE sets the industry standard for energy efficiency, computational density, and bandwidth.
System Highlights:
- Massive reductions in TCO: uses 1/2 the power and takes 1/3 the space of the best-in-class legacy server
- Industry-leading networking bandwidth: delivers 10 Gigabits of network bandwidth to each socket — an industry-leading 2.5 Gigabits per core
- Plug-and-play deployment: runs off-the-shelf OS, hypervisiors, and applications
- Eliminates expensive TOR switches and load balancers
- Reduces the number of end-of-row and core switches
- Guarantees performance and security for cloud deployments
Collapsing an Entire Rack into a Single 10 RU System
Three key SeaMicro technologies transform the computational landscape. The first, named SeaMicro I/O VT, eliminates all components from the motherboard except three—the SeaMicro Freedom™ ASIC, the CPU, and DRAM. SeaMicro’s second technology, named SeaMicro TIO™, reduces the power consumed by the CPU by consolidating and powering down unused functions. Together, SeaMicro I/O VT and TIO allow SeaMicro to reduce dramatically the size and power of the motherboard. SeaMicro’s third technology, the Freedom Supercomputer Fabric, links together these mini-motherboards with massive bandwidth and low latency to create a 1.28-Terabit supercomputing fabric.
Redundancy and Reliability
The SeaMicro SM10000-XE implements redundancy in hardware and in software and contains no single point of failure. On the hardware side, all subsystems are redundant and are hot-swappable, including computional cards, disk, network uplink cards, power supplies, and fans. On the software side, the SM10000-XE can run fully redundant control-plane software in active or standby mode on two separate management cards. In the event of a failure, standby software assumes the responsibility of managing the system without manual intervention. SeaMicro software also manages the Freedom™ Supercomputer Fabric and routes traffic around failure using multiple alternative fabric paths (path redundancy).
SeaMicro’s modular management software provides process isolation and modularity, with
each major process operating in its own address space—thereby increasing system availability and reliability.
Benefits
Reduction in Operating Expense from Best-in-Industry Energy Efficiency. Power is the single largest operating cost in the data center, often in excess of 30 percent of total operating expenses (Op Ex). Google, for example, reports that over a server’s life, the power bill exceeds its purchase price. The SM10000-XE is the most energy-efficient Xeon server ever built. It uses 1/2 the power to do the same work as the best-in-class competition.
Reduction in Operating Expense from Saved Space. Space is the second largest Op Ex line item. The SM10000-XE delivers three times the computations per rack unit of the best-in-class competitor, reducing the dollars spent on space by 66%. The SM10000-XE packs 256 Xeon cores into a 10-RU system, up to 1,024 Xeon cores in a single rack.
High Networking Bandwidth That Eliminates Stranded Computations. The bandwidth of 1.28 Terabits per second provided by the Freedom Supercomputer Fabric allocates 10 Gigabits of bandwidth to each of the 64 Xeon processors — 2.5 Gigabits per core. This is more than 12 times the bandwidth provided by the leading competitors. The extra bandwidth frees the processors to perform at peak levels without network bottlenecks that limit computational performance.
Massive Input/Output (I/O) Bandwidth. The SM10000-XE also supports up to 16 10-GbE uplinks, providing 160 Gbps bandwidth into the system and making the SM10000-XE an ideal platform for bandwidth-hungry applications seen in Web, streaming-video, and CDN applications.
Data-Center Design Simplified with Integrated Architecture. The SM10000-XE integrates 64 Sandy Bridge–class Xeon CPUs, two top-of-rack switches, two terminal servers, and a load balancer into a single 10-RU system. Eliminating these discrete devices reduces capital expense, simplifies infrastructure, and reduces management costs.
Hardware-Enforced Security and Performance for Cloud Deployments. Unlike traditional 1-RU or 2-RU servers, the SM10000-XE can guarantee performance and security at the hardware level—an important feature for cloud and managed hosting providers. When CPUs are shared across multiple users, as they are in today’s cloud environments, contention can limit performance and open security holes. SeaMicro’s architecture provides dedicated, right-sized computing units and the ability to enforce security rules and guarantee performance in hardware. In this way, Ethernet interfaces and disk interfaces can be securely tied to particular computing units including virtual machines, guaranteeing performance and security.
High Availability and Remote Monitoring at Low Cost for Clouds. The SM10000-XE implements a unique feature called the “spare server.” This enables customers to maintain a single shared-spare CPU to back up the 63 other CPUs in the system. If a failure occurs on an active CPU, the network parameters (IP address and MAC) and actual physical disks of the failed CPU are seamlessly moved, via remote management, to the shared-spare CPU, thereby allowing application services to be restored in record time.
Flexibility in the Ratio of Computational Uplink Bandwidth and Storage Capacity. The SeaMicro architecture gives customers the freedom to choose the amount of disk and uplink bandwidth they desire. Configure a system with no disk, some disk, or many disks. Configure a system from 8 1-Gb Ethernet uplinks to 16 10-Gb Ethernet uplinks. Tailor the ratio for your specific requirements of computation to disk and computation to I/O. For example, a Web server with a requirement for small data storage and uplink bandwidth can be configured with minimal disk space (reducing the need for expensive, failure-prone disks) and small network-uplink bandwidth, while a CDN system requiring high disk and network bandwidth could be configured with up to 16 10-GbE network uplinks and dedicated high-capacity disks and SSDs.
Seamless Addition of Disk and I/O Bandwidth. In a SM10000-XE, network-uplink bandwidth can be upgraded while in use without any system downtime. Uplink ports in increments of 2 10-GbE or 8 1-GbE can be inserted without powering down the system. Similarly, disk can be added as needed.
SAN-like Disk Management at DAS prices. Any CPU can be configured with multiple virtual disks, allowing better amortization of disks over computational resources. A virtual disk can be as large as a physical disk, or it can be a slice of a physical disk. A single physical disk can be portioned into multiple virtual disks, with each virtual disk allocated to a different CPU. Conversely, a single virtual disk can be shared across multiple CPUs in read-only mode, providing a large shared data cache. Sharing of a virtual disk enables users to store or update common data, such as operating systems, application software, and data cache, once for
an entire system.
Accelerated Deployment: Tight packaging of CPUs in a 10-RU appliance allows for simple installation. The steps needed to install a SeaMicro system are as follows:
(1) install the SeaMicro hardware into the rack;
(2) connect up to 4 power-supply cables to a power supply source;
(3) connect uplink cables from the SeaMicro appliance to the core switch;
(4) connect the management Ethernet cable to a management switch;
(5) configure the management boot parameters for all servers and uplink interfaces just
one time using the SeaMicro management-user interface.
Simplified Operations: All of the 64 sockets in a SeaMicro SM10000-XE can be managed remotely using SeaMicro’s redundant management infrastructure. A single management API can control CPU reset and power on/off; installation of new BIOS, operating system, and application software; dynamic modification of load-balancer capacity; and system-wide performance monitoring and troubleshooting. The SM10000-XE provides a rich suite of management API including XEN XML API, traditional CLI, SNMP, IPMI, and syslog. Using a rich suite of management API, customers can seamlessly integrate the SM10000-XE into existing operational service systems.
The SeaMicro SM10000 Family System Overview
This system overview provides a summary of the architecture and primary system components in the SM10000 family.
SeaMicro Technology Overview
The Technology Overview provides a brief summary of the major technology building blocks in the SeaMicro SM10000 family.
The SeaMicro SM10000 Redundancy and Reliability Overview
This white paper provides a summary of the resiliency characteristics of the SeaMicro SM10000 Server.
Report to Congress on Server and Data Center Energy Efficiency
This is an Environmental Protection Agency produced report to congress on the power problems and inefficiencies in the data center.
The EPA reports among other findings that Volume Servers are responsible for more than 1% of the power consumed in the United States and consume more than seven times the power of the second leading consumer of power—networking equipment.
Power Provisioning for a Warehouse-sized Computer
This paper written by engineers at Google focuses on the issue of power provisioning for large-scale Internet services. It was published in the Proceedings of the ACM International Symposium on Computer Architecture, San Diego, CA, June, 2007. Interestingly, they point out that the cost of power used by servers can exceed the purchase price of servers.





