The computer industry is on the major transition due to the server consolidation and virtualization products on the market. Cisco is commonly known for networking components like switches, routers and IP phones. But in 2009 , they have entered in to the X86 server market on their own style with data centre virtualization and consolidation methods.
There was a small history behind the Cisco UCS X86 servers.
- In 2008, VMware made a partnership with Cisco and the outcome was Cisco Nexus 1000V, a distributed virtual software switch which can be integrated with VMware vpShere.
- In 2009 , IBM, HP and Dell hardware vendors were leading in the blade servers market.
- At the same time , VMware also started capturing the server market with rock solid virtualization products.
- Cisco have proposed to IBM , Dell and HP to bring the blade server’s control plane to the Cisco UCS Manager to integrate the more networking features to the blades. This is nothing but managing the whole blade chassis from UCS manager.But they didn’t accept it.
- So Cisco has released their own X86 servers in the market with the brand name of Cisco Unified Computing System (UCS) in 2009.
- As a result of Cisco UCS success, DELL and HP are trying to do that what Cisco has done 5 years back.
What Cisco has done differently with UCS B-Servers blade servers ?
In HP,IBM and Dell blade servers are managed from chassis. But here the case is different . Cisco brought the control plane(L2 Layer) completely out from the chassis. The blade and chassis are managed by the Fabric interconnect which is nothing but a UCS manager.
Cisco UCS – B – Series Topology:
The below image shows the Cisco UCS chassis of back side & Front side.
1. FEX (Fabric Extended Module )or IOM (Inout/Output Module):
FEX extends the I/O fabric between the Cisco UCS Fabric Interconnects and the Cisco UCS B-Series Blade Server Chassis. The fabric extender is like a distributed line card, it does not perform any switching and is managed as an extension of the fabric interconnects. FEX manages the entire chassis which includes the blades, fans and power supply with help of Fabric interconnect. So, separate chassis management modules are not required.
FEX second generation(Cisco UCS 2200 series) contains 8 x 10Gigabit up links ports (chassis to FI connectivity) and it has thirty-two 10 Gigabit Ethernet ports connected through the midplane to each half-width slot in the chassis. Normally ,we will be configuring the FEX with pairs of redundancy. So it can provide up to 160 Gbps I/O to Chassis. (8x10Gbps + 8x10Gbps = 160Gbps.)
[box type=”info” align=”” class=”” width=””]Per Half Blade = 1 x 10Gbps (FEX 1) + 1x10Gbps (FEX 2) = 20Gbps of LAN/SAN traffic.
Full Blade = 2 x 10Gbps (FEX 1) + 2x10Gbps (FEX 2) = 40Gbps of LAN/SAN traffic.[/box]
Blades are connected internally with FEX in the chassis. So there is no cabling required between blades and FEX.
2. What is Fabric Interconnect (FI) ?
The Fabric interconnect is the heart of Cisco UCS systems. FI is provides the LAN/SAN Connectivity, Management to the chassis. In traditional blade systems , LAN & SAN network will be going to its pear part directly from chassis. Since all the Cisco blades are using mezzanine cards, both SAN & LAN traffic will be carried in to single medium till Fabric interconnect. From there traffic will be routed to its pear. FI runs on NX-OS and its nothing but a Nexus switch. UCS manager is the software which is installed on Fabric interconnect to manage the whole UCS infrastructure.
3. Do you would like to know that how the mezzanine card works ? Please Check out here.
We have seen the different components which are used in the Cisco B-series computing. Hope you got a better idea about it . In the upcoming article, we will see that how to discover the chassis from the UCS manager (FI)
Thank you for visiting UnixArena.
Leave a Reply