The IBM® System Networking RackSwitch G8264 delivers exceptional performance, being both lossless and low latency. In addition, the G8264 also delivers excellent cost savings as you consider acquisition costs, energy costs, operational expenses, and ease of use and management. It has a feature-rich design with key virtualization features such as CEE/FCoE, high availability, and enterprise class Layer 2 and Layer 3 functionalities.
The IBM RackSwitch G8264 supports both 10-Gb and 40-Gb Ethernet. It is suggested for clients who are using 10 Gb today with plans to include 40 Gb uplinks in the future. With 1.2 Tbps throughput, the G8264 provides massive scalability and low latency ideal for latency-sensitive applications such as high-performance computing clusters and financial applications.
IBM® System Networking RackSwitch™ G8264 leveraging 10Gb SFP+ and 40Gb QSFP+ Ethernet technology is specifically designed for the data center. It is ideal for today's big data, cloud, and optimized workload solutions. It is an enterprise class Layer 2 and layer 3 full featured switch that delivers line-rate, high-bandwidth switching, filtering, and traffic queuing without delaying data. Large data-center grade buffers help keep traffic moving, while the redundant power and fans along with numerous high availability features help provide high availability for business sensitive traffic.
The RackSwitch G8264, shown in Figure 1, is ideal for latency sensitive applications such as high performance computing clusters and financial applications. In addition to the 10-Gb Ethernet (GbE) and 40-GbE connections, the G8264 also has the capability for legacy 1-GbE connections. The G8264 supports the newest protocols including Data Center Bridging/Converged Enhanced Ethernet (DCB/CEE) for support of Fibre Channel over Ethernet (FCoE) in addition to iSCSI and network-attached storage (NAS).
Figure 1. IBM System Networking RackSwitch G8264
The G8264 also supports IBM Virtual Fabric, which helps clients significantly reduce cost and complexity related to I/O requirements of many virtualization deployments. Virtual Fabric helps reduce the number of multiple I/O adapters to a single dual-port 10 G adapter and reduces the number of cables and required upstream switch ports. By using Virtual Fabric, you can carve a dual-port 10 G server adapter into 8 virtual network interface cards (vNICs) and create dedicated virtual pipes between the adapter and switch for optimal performance, higher availability, and improved security. With Virtual Fabric IT, you can make dynamic changes and allocate bandwidth per vNIC so that you can adjust it over time without downtime.
The RackSwitch G8264 is the first 10 GbE switch to offer benefits of OpenFlow. OpenFlow is the new open application programming interface (API) that enables the network administrator to easily configure and manage virtual networks that control traffic on a “per-flow” basis. It creates multiple independent virtual networks and related policies without dealing with the complexities of the underlying physical network and protocols. The G8264 is also the recommended switch to use with the network controller that is compliant with OpenFlow, which is the IBM Programmable Network Controller.
Did you know?
- The RackSwitch G8264 supports several types of configurations: 1-Gb, 10-Gb, and 40-Gb, vNIC, Converged Enhanced Ethernet (CEE/DCB), NAS, and iSCSI.
- The RackSwitch G8264 with SFP+ technology is the ideal solution for clients who are looking for low latency and low power consumption.
- The IBM RackSwitch G8264 supports stacking for up to eight switches using a single switch image and configuration file that shares one IP address and one management interface.
- The RackSwitch supports Data Center Bridging (DCB), the group of protocols from IEEE that provide lossless Ethernet and that allow clients to reduce the costs of implementing either:
- NAS or iSCSI convergence and priority-based flow control
- FCoE by aggregating switch ports before connecting to more costly upstream Fibre Channel gateway devices.
- IBM is a leader in network virtualization, offering the on-switch IBM VMready® software that reduces the complexity of managing virtual machines (VMs) in the network. For more information, see IBM VMready at:
- The switch is OpenFlow enabled, which can make user-controlled virtual networks easy to create, optimize performance dynamically, and minimize complexity when used with IBM Programmable Network Controller. For more information, see IBM Programmable Network Controller at:
- The G8264 plays a vital role with IBM Virtual Fabric, which reduces costs and complexity in environments where four or more NICs are needed per server. An example is virtualization, where clients often need up to eight NICs per server. For more information, see IBM Virtual Fabric at:
Part number information
Table 1 lists the part numbers for ordering IBM System x® and IBM Power Systems™ modules and options. Products are ordered through IBM System x and IBM Storage and Power ordering systems.
Note: Transceivers are not included in the purchase of the switch and all transceivers require using the appropriate cables.
Table 1. IBM part numbers and feature codes for ordering (part 1)
|Description||System x part number||Power Systems feature code|
|IBM System Networking RackSwitch G8264 |
|IBM System Networking RackSwitch G8264|
|1 Gb options|
|IBM SFP RJ45 Transceiver||81Y1618||EB29|
|IBM SFP SX Transceiver||81Y1622||EB2A|
|IBM SFP LX Transceiver||90Y9424||ECB8|
|0.6 m Blue Cat5e Cable||40K5679||ECB0|
|1.5 m Blue Cat5e Cable||40K8785||ECB2|
|3 m Blue Cat5e Cable||40K5581||1111|
|10 m Blue Cat5e Cable||40K8927||1112|
|25 m Blue Cat5e Cable||40K8930||1113|
Table 1. IBM part numbers and feature codes for ordering (part 2)
|Description||IBM System x part number||Power Systems feature code|
|10 Gb options|
|IBM SFP+ SR Transceiver||46C3447||EB28|
|IBM SFP+ LR Transceiver||90Y9412||ECB9|
|IBM SFP+ ER Transceiver||90Y9415||ECBA|
|0.5 IBM Passive DAC SFP+ Cable*||00D6288||ECBG|
|1 m IBM Passive DAC SFP+ Cable*||90Y9427||ECB4|
|1.5 m IBM Passive DAC SFP+ Cable||00AY764|
|2 m IBM Passive DAC SFP+ Cable||00AY765|
|3 m IBM Passive DAC SFP+ Cable*||90Y9430||ECB5|
|5 m IBM Passive DAC SFP+ Cable*||90Y9433||ECB6|
|7 m IBM Passive DAC SFP+ Cable*||00D6151||ECBH|
|1 m IBM Active DAC SFP+ Cable||95Y0323|
|3 m IBM Active DAC SFP+ Cable||95Y0326|
|5 m IBM Active DAC SFP+ Cable||95Y0329|
|1m 10GbE Cable SFP+ Active Twinax||EN01|
|3m 10GbE Cable SFP+ Active Twinax||EN02|
|5m 10GbE Cable SFP+ Active Twinax||EN03|
Table 1. IBM part numbers and feature codes for ordering (part 3)
|40 Gb options|
|IBM QSFP+ SR4 Transceiver||49Y7884||EB27|
|1 m IBM QSFP+ DAC Break Out Cable*||49Y7886||EB24|
|3 m IBM QSFP+ DAC Break Out Cable*||49Y7887||EB25|
|5 m IBM QSFP+ DAC Break Out Cable*||49Y7888||EB26|
|1 m IBM QSFP+-to-QSFP+ Cable*||49Y7890||EB2B|
|3 m IBM QSFP+-to-QSFP+ Cable*||49Y7891||EB2H|
|5 m IBM QSFP+-to-QSFP+ Cable*||00D5810||ECBN|
|7 m IBM QSFP+-to-QSFP+ Cable*||00D5813||ECBP|
Table 1. IBM part numbers and feature codes for ordering (part 4)
|Miscellaneous options||System x part number||Power Systems feature code|
|IBM System Networking Adjustable 19" 4 Post Rail Kit||00D6185||EU27|
|IBM System Networking Recessed 19" 4 Post Rail Kit (NeXtScale)||00CG089|
|Front-to-rear, hot-swappable power supply spare||49Y7938|
|Rear-to-front, not swappable power supply spare||49Y7937|
|Front-to-rear, hot-swappable fan assembly spare||49Y7939|
|Rear-to-front, hot-swappable fan assembly spare||88Y6026|
|Console cable kit spare||90Y9462||EUC4|
With the flexibility of the IBM switch, you can take advantage of the following technologies required for multiple environments:
- For 1 Gb links, you can use SFP transceivers plus LC-to-LC fiber cables.
- For 10 Gb, you can use the following items:
- IBM Passive/Active Direct Attach Copper SFP+ Cable
- SFP+ transceivers (SR, LR, or ER) plus LC-to-LC fiber cables for longer distances
- For 40 Gb links, you can use QSFP to QSFP cables up to 7m, or QSFP transceivers and MTP cables for longer distances.
For the options and their supported distances, see Figure 2.
Figure 2. Options and their supported distances
The module part numbers include the following items:
- One IBM System Networking RackSwitch G8264
- Generic Rail Mount Kit (2-post)
- IBM serial cable
Power cords, listed in Table 2, are shipped separately at an additional charge.
Table 2. Power cords and descriptions: IBM System x
|39Y7917||Power Cord Europe AC plug 10A/250 V; OPT|
|39Y7918||Power Cord Europe (Denmark) AC plug 10A/250 V; OPT|
|39Y7919||Power Cord Europe (Switzerland) AC plug 10A/250 V; OPT|
|39Y7920||Power Cord Israel AC plug 10A/250 V; OPT|
|39Y7922||Power Cord South Africa AC plug 10A/250 V; OPT|
|39Y7923||Power Cord UK AC plug 13A/250 V; OPT|
|39Y7924||Power Cord Australia AC plug 10A/250 V; OPT|
|39Y7925||Power Cord Korea AC plug 10A/250 V; OPT|
|46M2593||Power Cord Japan AC plug 10A/125 V; OPT|
|39Y7927||Power Cord India AC plug 10A/250 V; OPT|
|39Y7928||Power Cord China AC plug 16A/250 V; OPT|
|39Y7929||Power Cord Brazil AC plug 16A/250 V; OPT|
|39Y7930||Power Cord Uruguay/Argentina AC plug 16A/250 V; OPT|
|46M2592||Power Cord US/Canada AC plug 10A/125 V; OPT|
Line cords, descriptions, and feature codes are listed in Table 3.
Table 3. Line cords description and IBM Power Systems and Storage feature codes
|Power Cord 4.3m (14-ft), Drawer to Wall/IBM PDU (250V/10A)||6458|
|Power Cord 4.3m (14-ft), Drawer to Wall/OEM PDU (250V/15A) U. S.||6469|
|Power Cord 1.8m (6-ft), Drawer to Wall (125V/15A)||6470|
|Power Cord 2.7m (9-ft), Drawer to Wall/OEM PDU (125V/15A)||6471|
|Power Cord 2.7m (9-ft), Drawer to Wall/OEM PDU (250V/16A)||6472|
|Power Cord 2.7m (9-ft), Drawer to Wall/OEM PDU (250V/10A)||6473|
|Power Cord 2.7m (9-ft), Drawer to Wall/OEM PDU, (250V/13A)||6474|
|Power Cord 2.7m (9-ft), Drawer to Wall/OEM PDU, (250V/16A)||6475|
|Power Cord 2.7m (9-ft), Drawer to Wall/OEM PDU, (250V/10A)||6476|
|Power Cord 2.7m (9-ft), Drawer to Wall/OEM PDU, (250V/16A)||6477|
|Power Cord 2.7m (9-ft), Drawer to Wall/OEM PDU, (125V/15A or 250V/10A)||6488|
|Power Cord 2.7m (9-ft), Drawer to Wall/OEM PDU, (250V/10A)||6493|
|Power Cord 2.7m (9-ft), Drawer to Wall/OEM PDU, (250V/10A)||6494|
|Power Cord 4.3m (14-ft), Drawer to Wall/OEM PDU (125V/15A)||6660|
|Power Cord 2.8m (9.2-ft), Drawer to Wall/IBM PDU, (250V/ 10A)||6665|
|Power Cord 2.7m (9-ft), Dawer to Wall/OEM PDU, (250V/10A)||6680|
|Power Cord 2.7m (9-ft), Drawer to Wall/OEM PDU, (250V/10A)||6671|
|Power Cord 1.5m (5-ft), Drawer to Wall/OEM PDU, (250V/10A)||6672|
The IBM System Networking RackSwitch G8264 offers the following benefits:
- High performance
The 10 Gb/40 Gb switch provides the best combination of low latency, non-blocking line-rate switching, and ease of management. It has a throughput of 1.28 Tbps.
- Lower power and better cooling
The IBM RackSwitch G8264 uses as little as 330 W of power, which is a fraction of the power consumption of most competitive offerings. Unlike side-cooled switches, which can cause heat recirculation and reliability concerns, the front-to-rear or rear-to-front cooling design of the G8264 reduces data center air conditioning costs by having airflow match the servers in the rack. In addition, variable speed fans help to automatically reduce power consumption.
With the G8264, a single switch image and configuration file can be used for up to eight switches, sharing only one IP address and one management interface.
- Virtual Fabric
The G8264 can help customers address I/O requirements for multiple NICs while reducing cost and complexity. Virtual Fabric for IBM allows you to carve a physical NIC into multiple virtual NICs (between 2 - 8 vNIC) and to create a virtual pipe between the adapter and the switch for improved performance, availability, and security.
- VM-Aware Networking
VMready software on the switch simplifies configuration and improves security in virtualized environments. VMready automatically detects virtual machine movement between physical servers and instantly reconfigures each VM’s network policies across VLANs to keep the network up and running without interrupting traffic or impacting performance. VMready works with all leading VM providers, such as VMware, Citrix, Xen, IBM PowerVM®, and Microsoft.
- Layer 3 functionality
The G8264 includes Layer 3 functionality, which provides security and performance benefits, as inter-VLAN traffic stays within the switch. This switch also provides the full range of Layer 3 protocols from static routes for technologies such as Open Shortest Path First (OSPF) and Border Gateway Protocol (BGP) for enterprise customers.
- Seamless interoperability
IBM switches perform seamlessly with other vendors' upstream switches.
- Fault tolerance
These IBM switches learn alternate routes automatically and perform faster convergence in the unlikely case of a link, switch, or power failure. The switch uses proven technologies like L2 trunk failover, advanced VLAN-based failover, VRRP, and HotLink.
The G8264 supports IGMP Snooping v1, v2, and v3 with 2 K IGMP groups. It also supports Protocol Independent Multicast (PIM), such as PIM Sparse Mode or PIM Dense Mode.
- Converged fabric
The IBM switch supports CEE and connectivity to FCoE gateways. CEE helps enable clients to combine storage, messaging traffic, VoIP, video, and other data on a common data center Ethernet infrastructure. FCoE helps enable highly efficient block storage over Ethernet for consolidating server network connectivity. As a result, clients can deploy a single server interface for multiple data types, which can simplify both deployment and management of server network connectivity, while maintaining the high availability and robustness required for storage transactions.
- OpenFlow enabled
The RackSwitch G8264 is the first 10 GbE switch to offer benefits of OpenFlow. OpenFlow is the new open API that enables the network administrator to easily configure and manage virtual networks that control traffic on a “per-flow” basis. It creates multiple independent virtual networks and related policies without dealing with the complexities of the underlying physical network and protocols. The G8264 is also the ideal switch to use with our OpenFlow compliant network controller, the IBM Programmable Network Controller.
- Transparent networking capability
With a simple configuration change to Easy Connect Mode, the RackSwitch G8264 becomes a transparent network device, invisible to the core, eliminating network administration concerns of Spanning Tree Protocol configuration/interoperability, VLAN assignments and avoids any possible loops.
By emulating a host NIC to the data center core, it accelerates the provisioning of VMs by eliminating the need to configure the typical access switch parameters.
Table 4. Comparison of the IBM RackSwitch G8264 and IBM RackSwitch G8264T features
IBM RackSwitch G8264
IBM RackSwitch G8264T
|Port types||Based on SFP+||Based on 10GBase-T|
|Latency||Lower latency - 880 nanosecond||3.2 microseconds|
|Typical power||Lower power - 330 watts||Competitive versus other 10GBase-T - 385W|
|Copper cable length||Up to 7m||Up to 100m|
|Total distance||Up to 40KM with Transceiver||Limited to 100m|
|TCO||Less than other SFP+||Lower TCO – for 100m and less|
(based on adapter, cables and switch)
The IBM RackSwitch G8264 includes an IBM standard 3-year hardware warranty with Next Business Day (NBD), 9x5, Customer Replaceable Unit (CRU) warranty service. IBM also includes a 3-year software warranty. IBM also reserves the right to provide new features as part of future software releases. Software Upgrade Entitlement is based on the switch’s warranty or post warranty extension and service contracts.
Features and specifications
Performance, hardware, and software features are described in the following lists.
- 100% Line rate performance
- 1280 Gbps non-blocking switching throughput (full duplex)
- Latency: 880 ns
- 960 Mpps
IBM RackSwitch G8264 has the following hardware features:
- IBM System Networking RackSwitch G8264 Rear-to-Front Versions
For ports located in the rear of the rack matching System x, IBM BladeCenter® and Power Systems designs
- IBM System Networking RackSwitch G8264 Front-to-Rear Versions
For ports located in the front of the rack matching airflow of IBM iDataPlex® ® and NeXtScale systems
- IBM System Networking RackSwitch G8264 Rear-to-Front Versions
- Interface options
- Forty-eight SFP/SFP+ ports (1/10 GbE)
- Four QSFP+ ports (40 GbE)
- Up to 64 10 Gb SFP+ connections with optional breakout cable
- One 10/100/1000 Ethernet RJ45 port for out-of-band management
- One USB port for mass storage device connection
- One mini-USB Console port for serial access
- Server-like port orientations, enabling short and simple cabling
- 17.3 inches wide, 19 inches deep, 1 RU high
- 10.5 kg (23 lb)
- Rack Installation Kit
- Generic Rack Mount Kit (2-post)
- Can be mounted vertically or horizontally
- System LEDs to indicate link status, fan, power, and stacking
- Front-to-rear or rear-to-front cooling
- Redundant variable hot-swap speed fans for reduced power draw
Figure 3 shows the rear view of the IBM RackSwitch G8264.
Figure 3. Rear view of IBM RackSwitch G8264
- Dual hot swap power modules, 50 - 60Hz, 100 - 240 VAC auto-switching per module
- Typical power consumption of 330 Watts
- Environmental specifications
- Temperature: Ambient operating: 0º C to + 40º C
- Relative humidity: Non-condensing, operating 10 to 90%
- Altitude: Operating 1,800 m (6,000 feet)
- Typical heat dissipation: 1127 BTU/hour
- Mean time between failures (MTBF):
- 165,990 hours at 40 °C
IBM RackSwitch G8264 has the following software features, as of 7.6:
- Wire Speed Filtering: Allow and Deny
- SSH v1, v2
- HTTPS Secure BBI
- Secure interface login and password
- MAC address move notification
- Shift B Boot menu (Password Recovery/ Factory Default)
- Port-based VLANs
- 4096 VLAN ids supported
- 2048 Active VLANs (802.1Q)
- 802.1x with Guest VLAN
- Private VLAN Edge
- FCoE/Lossless Ethernet
- 802.1 Data Center Bridging
- Priority Based Flow Control (PFC)
- Enhanced Transmission Selection (ETS)
- Data Center Bridge Exchange protocol (DCBX)
- FIP Snooping
- Fibre Channel over Ethernet
- Converged Enhanced Ethernet
- Static Trunks (EtherChannel)
- Configurable Trunk Hash algorithm
- Spanning Tree
- Multiple Spanning Tree (802.1 s)
- Rapid Spanning Tree (802.1 w)
- Fast Uplink Convergence
- BPDU guard
- Quality of Service
- QoS 802.1p (priority queues)
- DSCP remarking
- Routing protocols
- RIP v1/v2
- IPv6 OSPF v3 (RFC 5340)
- High availability
- Layer 2 failover
- Virtual Router Redundancy support (VRRP)
- IGMP Snooping v1, v2, and v3 with 2 K IGMP groups
- Protocol Independent Multicast (PIM) Sparse Mode/Dense Mode
- Port mirroring
- ACL-based mirroring
- sFlow Version 5
- VMready with VI API support
- vNIC MIB support for SNMP
- OpenFlow v1.3.10 (with IBM Networking OS 7.8)
- OpenFlow hybrid mode
- IBM System Networking Switch Center (SNSC)
- ISCLI (Cisco-like)
- Scriptable CLI
- Browser-based client or Telnet
- Standard protocols
- IPv4, IPv6
- SNMP v1, v2c, and v3
- Secondary NTP Support
- DHCP Client
- DHCP Relay
- 128 K MAC Table
- 9 K Jumbo Frames
- 802.3X Flow Control
- Upgrade firmware via serial or TFTP
- Dual software images
The following configurations are the most popular for using IBM RackSwitch G8264:
- Rack-optimized server aggregation 10-GbE attached rack servers
- Rack-optimized server aggregation logical design
- 2560 node gigabit network: Using next-generation 64 port 1U switches
- Virtual Fabric configuration with IBM System x servers using the Emulex Virtual Fabric Adapter
- IP Storage over 10 GbE: Changing the economics of storage
Rack-optimized server aggregation 10-GbE attached rack servers
The IBM RackSwitch G8264 is an ideal networking device for port aggregation, allowing the consolidation of one to three racks full of servers before connecting to an upstream switch.
- High-concentration of rack-optimized servers, for example:
- IBM System x 1U servers with a 10 Gb adapter installed
- Easily supports 40 servers per rack.
- Easily supports up to 48 servers or more, depending on uplink bandwidth required across two racks, including storage or UPS.
- IBM System x 2U servers with a 10 Gb adapter installed
- Most clients will have up to twenty 2U servers per rack.
- With the G8264 switch being able to support up to 64 SFP+ ports, a client can easily use a switch across three racks of severs, and more likely two for redundancy.
- IBM BladeCenter using any of the following modules in the chassis:
- IBM BladeCenter 1/10 Gb Uplink Ethernet Switch Module
With only two 10 Gb SFP+ ports per chassis, clients with larger data centers can easily support 24 chassis or more per RackSwitch G8264.
- IBM BladeCenter Virtual Fabric 10 Gb Switch Module
With support of up to ten 10 Gb SFP+ ports per switch, clients can easily support 4 - 6 chassis per RackSwitch, and if they are not concerned about oversubscription, they can support up to 24 or more per RackSwitch G8264.
- 10 Gb Ethernet Pass-Thru Module for IBM BladeCenter
For many clients looking to use a pass-through module, upstream costs can be expensive, especially if you are connecting to a core switch and are looking at up to fourteen 10 Gb ports per chassis. The G8264 provides a much more cost-effective alternative and can support up to four fully populated IBM BladeCenter H chassis per RackSwitch. This alternative is also much better for larger BladeCenter H environments compared to a smaller 24-port switch.
- IBM BladeCenter 1/10 Gb Uplink Ethernet Switch Module
- IBM System x 1U servers with a 10 Gb adapter installed
- Low-profile, high-performance, 48/64-port 10-GbE switch needed to perform aggregation function per rack. Figure 4 shows port aggregation, and Table 5 lists the features and benefits.
Figure 4. Port aggregation
Table 5. Features and benefits
|Line-rate, non-blocking, all 64-ports||Supports massive compute and virtualization workloads|
|Deterministically low latency||Faster application response times|
|Support for IGMP Snooping and L3 forwarding||Enables high-bandwidth, low-latency multicast applications|
|Standards-based Layer 2/3 protocols; industry standard CLI||Interoperates with existing network; no learning curve|
Rack-optimized server aggregation logical design
The design goal is to have the IBM RackSwitch G8264 (10 GbE) switch at the distribution layer and the IBM RackSwitch G8000/G8052 (1 GbE) at Edge/Access.
The G8264 is also ideal at the distribution layer when you are looking to aggregate connections for access layer switches like the G8000 or G8052 which support forty-eight 1-Gb ports to the servers and four 10-Gb uplinks. In either scenario, the G8264 provides exceptional value by providing the following support:
- Up to 12 G8000/G8052 switches (576 servers) when using the four 10 Gb uplinks per G8000/G8052, while still having the four 40 Gb uplinks from the G8264 to the core
- Up to 16 G8000/G8052 switches (768 servers) when using the four 10 Gb uplinks per G8000/G8052 and using 10 Gb SFP+ break-out cables in the four 40 Gb ports with no uplinks to a core
Keep in mind the following additional details:
- Logical configuration: Configure G8000/G8052 for Layer 2 and apply static routes for L3 forwarding
- Full Layer 2/3 Feature Set: STP, MSTP, RSTP, PVRST+; RIP v1/2, static routes, OSPF
- Security: 802.1X; RADIUS/TACACS+; Wire Speed ACLs, SSH v1, v2; HTTPS Secure BBI
- QoS: Up to eight queues per port, IEEE 802.1p and DiffServ prioritization
Figure 5 shows aggregation option 2.
Figure 5. Aggregation option 2
2560 node gigabit network: Using next-generation 64 port 1U switches
The following high-level configuration is an example of how a configuration might look if the G8052 is used as leaf switches and the G8264 is used as a spine switch.
- All active links
- Traffic load balanced across all links using Layer 3 ECMP
- 64 RackSwitch G8052 Leaf Switches for 2560 non-blocking 1 GbE ports
- 48x 10/100/1000 server-facing ports each
- 4x 10 GBASE-SR fiber uplinks each
- Four RackSwitch G8264 spine switches for non-blocking core
Figure 6 shows the 2560 node.
Figure 6. The 2560 node
Virtual Fabric configuration with IBM System x servers using the Emulex Virtual Fabric Adapter
The following example describes an IBM Virtual Fabric implementation:
- IBM Virtual Fabric for IBM System x is built using industry standards and is the ideal solution for customers that require more than two NICs per server. Over the past couple of years many clients have seen their I/O requirements per server increase anywhere from 4 to 6 to 8 or even more NICs per server, especially with the adoption of virtualization. What IBM Virtual Fabric does is provide clients with the ability to leverage a virtual NIC approach leveraging 10 Gb Ethernet technology to help reduce cost and complexity while also achieving better performance and more flexibility. It provides the following benefits:
- Reduced costs
Acquisition and operating by using fewer adapters, cables, and upstream switch ports.
- Reduced complexity
Fewer items to manage, based standards, making management easier with the addition of high availability and better security.
- Better performance
A significant amount more I/O bandwidth per server and lower latency than traditional 1-Gb Ethernet.
The ability to carve up a dual-port 10 Gb adapter into 4 - 8 virtual NICs and create virtual pipes between the adapter and switch for higher availability and security. Not only can customers dynamically allocate I/O bandwidth to each virtual NIC, Virtual Fabric also provides the ability to change those allocations as needed. For example, a customer might have one configuration for workload during the day and another during the evenings. Virtual Fabric has the ability to make those changes on the fly without downtime.
- Reduced costs
- IBM Virtual Fabric is designed in partnership with Emulex. This solution requires the following components:
- Emulex 10 Gb Virtual Fabric Adapter for IBM System x
For more details, see Emulex 10Gb Virtual Fabric Adapter for IBM System x, TIPS0762 at:
- IBM System Networking RackSwitch G8264
- Emulex 10 Gb Virtual Fabric Adapter for IBM System x
Figure 7. vNIC
IP Storage over 10 GbE: Changing the economics of storage
G8264 benefits for IP Storage applications are as follows:
- Lossless Ethernet Fabric
- Low latency ideal for HPC clusters, cloud, and financial applications
- Line-rate, high-bandwidth performance
- Low-power consumption with fewer components
- Low-cost, pay-as-you-grow 10 Gb/40 Gb storage network
Figure 8 shows the iSCSI.
Figure 8. iSCSI
For more information, see the following references:
- IBM Offering Information page (to search on announcement letters, sales manuals, or both):
On this page, enter G8264 for System x and iDataPlex, select the information type, and then click Search. On the next page, narrow your search results by geography and language.
- IBM System Networking RackSwitch G8264 product publications
- Application Guide
- Industry-Standard CLI Reference
- Browser-Based Interface (BBI) Quick Guide
- Menu-based CLI Command Reference
- IBM VMready
- IBM Programmable Network Controller
- IBM Virtual Fabric
This material has not been submitted to any formal IBM test and is published AS IS. It has not been the subject of rigorous review. IBM assumes no responsibility for its accuracy or completeness. The use of this information or the implementation of any of these techniques is a client responsibility and depends upon the client's ability to evaluate and integrate them into the client's operational environment.