The documentation set for this product strives to use bias-free language. For the purposes of this documentation set, bias-free is defined as language that does not imply discrimination based on age, disability, gender, racial identity, ethnic identity, sexual orientation, socioeconomic status, and intersectionality. Exceptions may be present in the documentation due to language that is hardcoded in the user interfaces of the product software, language used based on RFP documentation, or language that is used by a referenced third-party product. Learn more about how Cisco is using Inclusive Language.
Discusses downloading and managing firmware, upgrading through Auto Install, upgrading through service profiles, directly upgrading at endpoints using firmware auto sync, managing the capability catalog,
deployment scenarios, and troubleshooting.
Discusses the new licenses, registering Cisco UCS domain with Cisco UCS Central, power capping, server boot, server profiles, and server-related policies.
Discusses all aspects of management of UCS S-Series servers that are managed through Cisco UCS Manager.
Cisco Unified Computing System Overview
Cisco UCS has a unique architecture that integrates compute, data network access, and storage network access into a common
set of components under a single-pane-of-glass management interface.
Cisco UCS fuses access layer networking and servers. This high-performance, next-generation server system provides a data
center with a high degree of workload agility and scalability. The hardware and software components support Cisco's unified
fabric, which runs multiple types of data center traffic over a single converged network adapter.
Figure 1. Cisco Unified Computing System Architecture
Architectural Simplification
The simplified architecture of Cisco UCS reduces the number of required devices and centralizes switching resources. By eliminating switching inside a chassis, network
access-layer fragmentation is significantly reduced. Cisco UCS implements Cisco unified fabric within racks and groups of
racks, supporting Ethernet and Fibre Channel protocols over 10/25/40 Gigabit Cisco Data Center Ethernet and Fibre Channel
over Ethernet (FCoE) links. This radical simplification reduces the number of switches, cables, adapters, and management points
by up to two-thirds. All devices in a Cisco UCS domain remain under a single management domain, which remains highly available
through the use of redundant components.
High Availability
The management and data plane of Cisco UCS is designed for high availability and redundant access layer fabric interconnects. In addition, Cisco UCS supports existing
high availability and disaster recovery solutions for the data center, such as data replication and application-level clustering
technologies.
Scalability
A single Cisco UCS domain supports multiple chassis and their servers, all of which are administered through one Cisco UCS Manager. For more
detailed information about the scalability, speak to your Cisco representative.
Flexibility
A Cisco UCS domain allows you to quickly align computing resources in the data center with rapidly changing business requirements. This
built-in flexibility is determined by whether you choose to fully implement the stateless computing feature. Pools of servers
and other system resources can be applied as necessary to respond to workload fluctuations, support new applications, scale
existing software and business services, and accommodate both scheduled and unscheduled downtime. Server identity can be abstracted
into a mobile service profile that can be moved from server to server with minimal downtime and no need for additional network
configuration.
With this level of flexibility, you can quickly and easily scale server capacity without having to change the server identity
or reconfigure the server, LAN, or SAN. During a maintenance window, you can quickly do the following:
Deploy new servers to meet unexpected workload demand and rebalance resources and traffic.
Shut down an application, such as a database management system, on one server and then boot it up again on another server
with increased I/O capacity and memory resources.
Optimized for Server Virtualization
Cisco UCS has been optimized to implement VM-FEX technology. This technology provides improved support for server virtualization, including
better policy-based configuration and security, conformance with a company's operational model, and accommodation for VMware's
VMotion.
Unified
Fabric
With unified fabric,
multiple types of data center traffic can run over a single Data Center
Ethernet (DCE) network. Instead of having a series of different host bus
adapters (HBAs) and network interface cards (NICs) present in a server, unified
fabric uses a single converged network adapter. This type of adapter can carry
LAN and SAN traffic on the same cable.
Cisco UCS
uses Fibre Channel over Ethernet (FCoE) to carry Fibre Channel and Ethernet
traffic on the same physical Ethernet connection between the fabric
interconnect and the server. This connection terminates at a converged network
adapter on the server, and the unified fabric terminates on the uplink ports of
the fabric interconnect. On the core network, the LAN and SAN traffic remains
separated.
Cisco UCS
does not require that you implement unified fabric across the data center.
The converged network
adapter presents an Ethernet interface and Fibre Channel interface to the
operating system. At the server, the operating system is not aware of the FCoE
encapsulation because it sees a standard Fibre Channel HBA.
At the fabric
interconnect, the server-facing Ethernet port receives the Ethernet and Fibre
Channel traffic. The fabric interconnect (using Ethertype to differentiate the
frames) separates the two traffic types. Ethernet frames and Fibre Channel
frames are switched to their respective uplink interfaces.
Fibre Channel over
Ethernet
Cisco UCS
leverages Fibre Channel over Ethernet (FCoE) standard protocol to deliver Fibre
Channel. The upper Fibre Channel layers are unchanged, so the Fibre Channel
operational model is maintained. FCoE network management and configuration is
similar to a native Fibre Channel network.
FCoE encapsulates
Fibre Channel traffic over a physical Ethernet link. FCoE is encapsulated over
Ethernet with the use of a dedicated Ethertype, 0x8906, so that FCoE traffic
and standard Ethernet traffic can be carried on the same link. FCoE has been
standardized by the ANSI T11 Standards Committee.
Fibre Channel traffic
requires a lossless transport layer. Instead of the buffer-to-buffer credit
system used by native Fibre Channel, FCoE depends upon the Ethernet link to
implement lossless service.
Ethernet links on the
fabric interconnect provide two mechanisms to ensure lossless transport for
FCoE traffic:
Link-level flow
control
Priority flow
control
Link-Level Flow Control
IEEE 802.3x link-level flow control allows a congested receiver to
signal the endpoint to pause data transmission for a short time. This
link-level flow control pauses all traffic on the link.
The transmit and receive directions are separately configurable. By
default, link-level flow control is disabled for both directions.
On each Ethernet interface, the fabric interconnect can enable either
priority flow control or link-level flow control (but not both).
When an interface on a Cisco UCS 6454 Fabric Interconnect has Priority Flow Control (PFC) admin configured as auto and Link-Level Flow Control (LLFC) admin configured as on, the PFC operation mode will be off and the LLFC operation mode will be on. On UCS 6300 Series and earlier Fabric Interconnects, the same configuration will result in the PFC operation mode being
on and the LLFC operation mode being off.
Priority Flow Control
The priority flow control (PFC) feature applies pause functionality to
specific classes of traffic on the Ethernet link. For example, PFC can provide
lossless service for the FCoE traffic, and best-effort service for the standard
Ethernet traffic. PFC can provide different levels of service to specific
classes of Ethernet traffic (using IEEE 802.1p traffic classes).
PFC decides whether to apply pause based on the IEEE 802.1p CoS value.
When the fabric interconnect enables PFC, it configures the connected adapter
to apply the pause functionality to packets with specific CoS values.
By default, the fabric interconnect negotiates to enable the PFC
capability. If the negotiation succeeds, PFC is enabled and link-level flow
control remains disabled (regardless of its configuration settings). If the PFC
negotiation fails, you can either force PFC to be enabled on the interface or
you can enable IEEE 802.x link-level flow control.
Multilayer Network
Design
When you design a
data center using a modular approach, the network is divided into three
functional layers: Core, Aggregation, and Access. These layers can be physical
or logical, and you can add and remove them without redesigning the entire data
center network.
Because of the
hierarchical topology of a modular design, the addressing is also simplified
within the data center network. Modularity implies isolation of building
blocks, which are separated from each other and communicate through specific
network connections between the blocks. Modular design provides easy control of
traffic flow and improved security. In other words, these blocks are
independent from each other; a change in one block does not affect other
blocks. Modularity also enables faster moves, adds, and changes (MACs) and
incremental changes in the network.
Modular networks are
scalable. Scalability allows for the considerable growth or reduction in the
size of a network without making drastic changes. Scalable data center network
design is achieved by using the principle of hierarchy and modularity.
Keep a network as
simple as possible. Modular designs are simple to design, configure, and
troubleshoot.
Access Layer—The access
layer is the first point of entry into the network for edge devices, end
stations, and servers. The Access layer grants user access to network devices
and provides connectivity to servers. The switches in the access layer are
connected to two separate distribution layer switches for redundancy. The data
center access layer provides Layer 2, Layer 3, and mainframe connectivity. The
design of the access layer varies, depending on whether you use Layer 2 or
Layer 3 access. The access layer in the data center is typically built at Layer
2, which allows better sharing of service devices across multiple servers. This
design also enables the use of Layer 2 clustering, which requires the servers
to be Layer 2 adjacent. With Layer 2 access, the default gateway, you can
configure for the servers at the aggregation layer.
Aggregation Layer—The
aggregation (or distribution) layer aggregates the uplinks from the access
layer to the data center core. This layer is the critical point for control and
application services. Security and application service devices (such as
load-balancing devices, SSL offloading devices, firewalls, and IPS devices) are
often deployed as modules in the aggregation layer. The aggregation layer
provides policy-based connectivity.
Core Layer—Also known as
backbone, the core layer provides high-speed packet switching, scalability and
high availability, and fast convergence. Implementing a data center core is a
best practice for large data centers. When you implement the core in an initial
data center design, it eases network expansion and avoids disruption to the
data center environment.
Use the following
criteria to determine whether a core solution is appropriate: The data center
typically connects to the campus core using Layer 3 links. The data center
network is summarized, and the core injects a default route into the data
center network.
Feedback