Architectural Comparison: Cisco UCS and the Dell FX2 Platform

Transcription

1 White Paper Architectural Comparison: Cisco UCS and the Dell FX2 Platform What You Will Learn Cisco Unified Computing System (Cisco UCS ) is more than a traditional blade solution. Unbound from traditional products and thinking, Cisco UCS is the first truly unified data center platform that combines industry-standard, x86-architecture servers with networking and storage access into a single system. This radically simplified solution is intelligent infrastructure that is automatically configured through integrated, model-based management to accelerate deployment of all your enterprise-class applications and services running in bare-metal, virtualized, and cloud-computing environments. Traditional blade architectures such as the Dell FX2 originally evolved from the idea of repackaging rack servers and switching as a smaller form factor. As density and performance increased, so did complexity. As a result, these traditional blade solutions have failed to deliver the promised consolidation. Cisco revolutionized blade servers through the unification of networking and management fabrics and delivers on the original vision of true consolidation of resources, time, and energy. This document shows you why Cisco and/or its affiliates. All rights reserved. This document is Cisco Public information.

2 Contents Approaches to Increasing Density... 4 Dell PowerEdge FX2 Enclosure... 4 Cisco UCS Blade Server Chassis... 4 Power and Cooling Comparison... 5 Networking Comparison... 6 Dell FX2 Enclosure with Pass-Through Networking... 6 Dell FX2 Enclosure with I/O Aggregator Switches... 7 Dell FX2 Enclosure with LAN and SAN... 7 Cisco UCS with SingleConnect Technology... 9 Management Comparison Management Cabling Management Complexity Simplified Management Performance and Choice of Processors Conclusion For More Information Cisco and/or its affiliates. All rights reserved. This document is Cisco Public information. Page 2

3 Architectural Comparison: Cisco UCS and the Dell FX2 Platform White Paper The promise of blade servers was to overcome the limitations and inefficiencies of rack-mount servers. Cisco UCS delivers on the promise. The vision for blade servers was that they would form a resource pool sharing a set of environmental, networking, and storage resources, rather than replicating all resources for all servers, as is done with rack-mount servers. In this way, blade servers would reduce both capital expenditures (CapEx) and operating expenses (OpEx). According to the book Blade Servers and Virtualization 1, Blade server systems offer a highly modular, prewired, ultrahigh-density form factor for servers in the data center. With a design that allows shared redundant components such as power and cooling, network switches, and management modules, blade servers help reduce the number of components in the data center as well as significantly increase the computing power per footprint. Blades and virtualization are key enablers for true unity computing, an environment in which components are added (provisioned) when they are needed, where they are needed, and only for as long as they are needed to give control back to businesses. Blade servers together with server virtualization were meant to create a set of computing resources that could be provisioned and reprovisioned as the business needed. Dell has tried to implement this vision twice: first with the Dell M1000e blade enclosure and now with the Dell FX2 architecture. Meanwhile, the Cisco Unified Computing System is the only blade system on the market today that delivers true 1 Blade Servers and Virtualization, Transforming Enterprise Computing While Cutting Costs, Barb Goldworm and Anne Skamarock, Wiley Publishing, Cisco and/or its affiliates. All rights reserved. This document is Cisco Public information. Page 3

4 unity computing while also giving you choice. This document describes the benefits of the blade server vision and how Dell and Cisco measure up to the vision. Approaches to Increasing Density The shortcomings of traditional blade architectures, including the Dell FX2 architecture, begin with the blade chassis itself. The rack-in-a-box blade architecture takes all the complexity of a rack top-of-rack (ToR) switches, management modules, redundant components and squeezes each component into every blade chassis. When you replicate this complexity multiple times per rack, you get even more complexity. In fact, you get far more complexity than a rack full of traditional servers with ToR switches. Both Dell s M1000e and FX2 platforms turn each chassis into an individual island that doesn t allow bandwidth to be shared outside each chassis. When you scale applications or blade server resources, you have to reconsider the way that all components in the chassis are connected. When you want to scale your applications, you must constantly evaluate whether and when a new switch module is necessary, what kind of I/O adapters are to be used on each affected blade, and whether or not recabling is required. Changing networking modules or updating blades to take advantage of new Intel architectures requires administrators to continually rethink and manage each island. Dell PowerEdge FX2 Enclosure Dell created the PowerEdge FX2 enclosure in a 2-rack-unit (2RU) form factor. In this design, sharing occurs between the devices installed in those two rack units, but not beyond. Each FX2 enclosure can support up to eight quarter-width blade servers, called server blocks. The enclosure can support four half-width server blocks, two full-width server blocks, or a combination of server blocks. The server blocks in each enclosure share power, cooling, management, and peripheral component interconnect (PCI) connectivity. The FX2 enclosure thus requires infrastructure repetition for every two units in each rack, including fans, power supply units, switching, cabling, and management. Dell currently offers three server blocks for the FX2 enclosure with Intel Xeon processors: a 2-socket quarter-width server block, a 2-socket half-width server block, and a 4-socket full-width server block. The examples used in this document employ the Dell PowerEdge FC630, the half-width blade server block. Cisco UCS Blade Server Chassis Each Cisco UCS chassis is 6RU and can support eight half-width blades, four full-width blades, two full-width double-height blades, or a combination of blade servers. Cisco currently offers five blade server products spanning all the form factors and ranging from 2 to 4 sockets. Within each Cisco UCS B-Series Blade Server Chassis, servers share power, cooling, and bandwidth resources. With a central point of connectivity and management, Cisco UCS is designed so that all chassis connected to the fabric interconnects comprise a single logical blade chassis that is logically centralized but physically distributed. Within each system, up to 160 blade servers (up to 20 chassis), plus their networking and management, are shared (Figure 1) Cisco and/or its affiliates. All rights reserved. This document is Cisco Public information. Page 4

5 Cisco UCS Management Cisco UCS 5108 Blade Server Chassis Cisco UCS B-Series Blade Servers Mission Critical Cisco UCS B460 M4 Cisco UCS B260 M4 Cisco UCS Fabric Interconnects Cisco UCS 6324 Fabric Interconnect Cisco UCS and Cisco Nexus Fabric Extenders Cisco UCS 2208XP Fabric Extender Enterprise Class Scale Out Cisco Cisco UCS B420 M4 Cisco and M4 Cisco UCS C-Series Rack Servers Cisco UCS Virtual Interface Cards Mezzanine and Modular LOM Form Factor PCIe Card Form Factor Cisco UCS 6296UP 96-Port Fabric Interconnect Cisco UCS 2204XP Fabric Extender Mission Critical Cisco UCS VIC Integrated into Cisco UCS M-Series Chassis Cisco UCS C460 M4 Cisco UCS 6248UP 48-Port Fabric Interconnect Optional Cisco Nexus 2232PP 10GE Fabric Extender Enterprise Class Cisco UCS C240 M4 Cisco UCS C240 M3 Cisco UCS C220 M4 Scale Out Cisco UCS C3160 Cisco UCS C22 M3 Cisco UCS C220 M3 Cisco UCS C24 M3 Cisco UCS M-Series Modular Servers Cisco UCS M4308 Modular Chassis Cisco UCS M142 Compute Cartridge Figure 1. Cisco UCS Configuration Options Power and Cooling Comparison With high-density computing, power and cooling often present a challenge. Often companies choose greater density over processor performance and high-memory configurations simply because, at those densities, it is impossible to keep the processors and memory within normal operating temperatures. Although Dell likes to position its FX2 platform as supporting extreme rack densities, the platform also requires significant redundancy and high power consumption. Every 2RU chassis requires two power supply units (PSUs), fans, chassis management resources, and switches. A typical PowerEdge FC630 configuration might use four half-width server blocks each with two Intel Xeon processor E v3 CPUs, twenty-four 16-GB DDR4 RDIMMs, a Broadcom K converged network adapter (CNA), eight fans, two switches, and two 300-GB disk drives. According to the Dell Energy Smart Solution Advisor (ESSA) power estimator, which historically underestimates power, this configuration requires nearly 2000 watts (W). That amounts to a kilowatt per rack unit: a huge requirement. Few data centers designed to support up to 10 kw of cooling per rack have sufficient power and cooling available to support a full rack of these servers Cisco and/or its affiliates. All rights reserved. This document is Cisco Public information. Page 5

6 For example, if you were to deploy only a half-rack of these chassis 10 chassis or 20RU you would need: 10 FX2 chassis 40 FC630 server blocks (20 cores each, 384 GB of memory, with 600 GB of disk space) 80 fans 20 switches 10 management control units 20 x 1000W = 20 kw You could easily exceed 20 kw of power, and that doesn t include the ToR switches required to aggregate the I/O connectivity from each of the FX2 enclosures. Compare this configuration to an equivalent Cisco UCS configuration. Each 6RU chassis can contain eight Cisco UCS B200 M4 Blade Servers, each with two Intel Xeon processor E v3 CPUs, twenty-four 16-GB DDR4 RDIMMs, two 300-GB disks, one Cisco UCS Virtual Interface Card (VIC) 1340 modular LAN on motherboard (mlom), and two Cisco UCS 2208 Fabric Extenders), averaging 625W per rack unit. To support five Cisco UCS chassis (32RU), you would need: 5 Cisco UCS 5108 Blade Server Chassis 40 Cisco UCS B200 M4 Blade Servers (20 cores each, 384 GB of memory, with 600 GB of disk space) 20 fans 10 Cisco UCS 2208 Fabric Extenders 30 x 625W = kw ToR switches are not included in either comparison. In the case of Cisco UCS, two Cisco UCS 6248UP 48-Port Fabric Interconnects would be required, consuming a total of only 2RU and 530W. Networking Comparison The Dell FX2 enclosure has no unified fabric. It uses a networking architecture with redundant pass-through modules or switches called I/O aggregators. What this approach means for you is a lot of ToR switch ports and a lot of cabling. Dell FX2 Enclosure with Pass-Through Networking To see the effects of this approach, consider an example with 10 Dell FX2 enclosures each populated with four PowerEdge FC630 blades. The networking requirements to configure just a LAN when using pass-through modules includes (as for all rack servers) two adapters per blade, or eight per FX2 enclosure, plus an adapter for management, for a total of nine cables per FX2 chassis (Figure 2). With 10 FX2 enclosures, the total requirements are 90 cables connecting to 90 ToR ports Cisco and/or its affiliates. All rights reserved. This document is Cisco Public information. Page 6

7 ToR Switches: Management Network Switch Production Network Switches 2 network connections to each server = 80 cables + 1 management network connection per enclosure = 10 cables Total of 90 cables and upstream switch ports 1 Management Module per Enclosure 2 Pass-Through Modules per Enclosure 10 Dell FX2 Enclosures Populated with 4 FC630 Server Blocks Each Figure 2. Using Pass-Through Modules, a Sample Configuration Requires 90 Cables and Switch Ports The worst-case scenario is an enclosure with eight Dell PowerEdge FC430 blades. This configuration requires 160 cables to maintain redundant connectivity to each server block. This number, plus a total of 10 management network cables, brings the maximum cabling requirement to 170 cables and upstream switch ports. Dell FX2 Enclosure with I/O Aggregator Switches If you use two I/O aggregators per chassis, the networking requirements (Figure 3) include two switches per FX2 enclosure with one chassis management module. To support redundant 10-Gbps connectivity per blade, each switch needs two network cables per chassis to the ToR switches plus one chassis management cable, for a total of five cables per chassis. With 10 FX2 enclosures, you will have total of 50 cables connecting to the ToR switches. In addition, you will have 20 in-chassis switches to manage. Dell FX2 Enclosure with LAN and SAN If you configure both a LAN and a SAN using Dell I/O aggregators (that support both LAN and SAN), you will need (Figure 4): two switches per chassis, two LAN connections per switch, at least one SAN connection per switch, and one connection for management, for a total of seven cables per chassis. With 10 FX2 enclosures, you thus need a total of 70 cables connecting to ToR SAN and LAN ports, and you will have 20 in-chassis switches to manage Cisco and/or its affiliates. All rights reserved. This document is Cisco Public information. Page 7

8 Architectural Comparison: Cisco UCS and the Dell FX2 Platform ToR Switches: Management Network Switch Production Network Switches 2 network connections to each I/O aggregator = 40 cables + 1 management network connection per enclosure = 10 cables Total of 50 cables and upstream switch ports 1 Management 2 I/O Aggregators Module per per Enclosure Enclosure 10 Dell FX2 Enclosures Populated with 4 FC630 Server Blocks Each Figure 3. Using I/O Aggregators, the Sample Configuration Requires 50 Cables and Switch Ports ToR Switches: Management Network Switch Production Network Switches Fibre Channel Switches 2 network connections to each I/O aggregator = 40 cables + 1 Fibre Channel connection to each I/O aggregator = 20 cables + 1 management network connection per enclosure = 10 cables Total of 70 cables and upstream switch ports 1 Management 2 I/O Aggregators Module per per Enclosure Enclosure 10 Dell FX2 Enclosure Populated with 4 FC630 Server Blocks Each Figure 4. Using Dell I/O Aggregators, the Sample Example Configuration Requires 70 Cables and Switch Ports to Support Fibre Channel Access to Storage 2015 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public information. Page 8

9 UCS C24 M3 Architectural Comparison: Cisco UCS and the Dell FX2 Platform Cisco UCS with SingleConnect Technology Cisco UCS is the only integrated system that reduces the number of hardware components and combines both blade and rack servers in a single unified fabric and management domain. Our approach eliminates management modules and switches in every chassis, reducing the cost of powering, cooling, configuring, managing, monitoring, and maintaining the infrastructure. Cisco UCS places all management functions and configuration information in the fully redundant and highly available Cisco UCS Manager. Cisco SingleConnect technology provides a way to connect all the devices and virtual machines in a single Cisco UCS domain. It provides a low-latency, highbandwidth 10 Gigabit Ethernet unified fabric that carries three networks in one: networks for IP, Fibre Channel over Ethernet (FCoE), and management traffic. The system s fabric extenders and VICs condense network layers so that only a single network hop is required for traffic from the server or virtual machines to the fabric interconnects. SingleConnect is implemented using industry standards and with performance accelerated by custom silicon (Figure 5). End-to-End I/O Architecture with Performance Accelerated by Cisco Silicon Cisco UCS 6200 Series Fabric Interconnects Cisco UCS and Nexus Fabric Extenders Cisco UCS Virtual Interface Cards (VICs) Physical Servers and Virtual Machines Blade and Rack Servers Virtual Machines Unified Crossbar Fabric Unified Port Controllers Multiplexer ASIC Cisco VIC ASIC Virtual NICs and HBAs Figure 5. Cisco UCS Cabling Simplicity of Cisco SingleConnect Technology Accelerated Through Custom Silicon Unified Fabric That Scales on Demand Cisco unified fabric helps you scale data centers easily, quickly, and efficiently without requiring a reevaluation of networking infrastructure every time you add new servers. With Cisco UCS, the network is established once, with no changes necessary as it scales to 160 servers per domain (or multiple domains of up to 6000 servers). Through aggregation of management and connectivity in the fabric interconnects, every server in the domain is automatically connected northbound to the LAN or SAN without time-consuming and risky reconfiguration at the chassis and server levels Cisco and/or its affiliates. All rights reserved. This document is Cisco Public information. Page 9

10 Unlike Dell s network, in which bandwidth choices are made at deployment time and which cannot be reconfigured without recabling the network, Cisco UCS network is unified and bandwidth is shared dynamically. This approach provides more effective use of bandwidth and more headroom in the event that an application needs to borrow bandwidth from other network modalities. Cisco UCS uses quality-of-service (QoS) controls to assign bandwidth priorities, and it enables your applications to dictate connectivity and bandwidth. Single Network Layer That Simplifies Networking Cisco UCS uses two low-power, zero-management Cisco UCS fabric extenders per chassis to forward traffic to two Cisco UCS fabric interconnects. You cable the fabric extenders to handle the desired amount of bandwidth from each chassis: from two to sixteen 10-Gbps cables per chassis. The fabric extenders configure all cables into a PortChannel that dynamically balances IP, storage, and management network traffic to the fabric interconnects. If you discover that you need more bandwidth as your application needs grow, all you need to do is add more cables, not reconfigure your entire network. From the beginning the bandwidth to each chassis is shared, so you need to size only for your expected peak usage across the entire chassis, not the peak usage for each blade. Shared Bandwidth Continuing the 40-server example, you can configure Cisco UCS with a total of 20 cables to provide up to 40 Gbps of burst capacity per server. Only 20 cables are needed to provide IP, FCoE, and management network connectivity to each blade server: considerably fewer than the 70 cables needed to provide equivalent capabilities in the Dell example, and with 34 fewer points of management (Figure 6). If you want to provide 10 Gbps of dedicated bandwidth per server, you can do so simply by increasing the number of cables to 40. This configuration supports 80 Gbps of bandwidth to each chassis, and because it is shared bandwidth, any server can burst up to 80 Gbps if you allow this in your QoS settings. The maximum connectivity to each blade chassis is 160 Gbps (using Cisco UCS 2208XP Fabric Extenders), which brings the maximum burst bandwidth to up to 160 Gbps per fullwidth blade server. Deterministic Latency Cisco UCS fabric interconnects have deterministic, low latency. Cisco UCS fabric interconnects centrally manage network traffic, passing network traffic in and out of the system. Traffic that is moving from one chassis to another in the same Cisco UCS domain does not need to exit the system, with packets sent through an external switch (see path A in Figure 7). This central connectivity demonstrates one of the ways in which Cisco UCS functions as a single virtual blade server chassis and provides the flexibility to place workloads anywhere in the system with consistent network performance. This capability is important because it simplifies placement of virtual machines in virtualized and cloud environments: regardless of location, virtual machines experience the same latency, removing the network as a placement constraint Cisco and/or its affiliates. All rights reserved. This document is Cisco Public information. Page 10

11 Architectural Comparison: Cisco UCS and the Dell FX2 Platform Cisco UCS Port Fabric Interconnects Gigabit Ethernet and FCoE Cisco UCS 2204XP Fabric Extenders Cisco UCS 5108 Blade Server Chassis with 8 Cisco UCS B200 M4 Blade Servers Total of 20 Cables to Provide 40 Gbps of Burst Capacity in IP, Fibre Channel, and Management Connectivity to Every Server Figure 6. Cisco UCS Requires Fewer Cables and Management Points Compared to the Dell FX2 Architecture When the goal is to reduce latency between servers, the best-case scenario for any vendor is a single network hop for data communicated between servers within the same chassis. The worst-case scenario is three network hops required to move data between servers in different chassis. With the Dell FX2 enclosure, traffic between each enclosure must first travel through the local switch module, through an external ToR switch, and back through the second enclosure s I/O aggregator, resulting in three network hops and greater latency. Although Dell refers to its mechanism as an I/O aggregator, it contains a switch application-specific integrated circuit (ASIC) that performs switching functions. Therefore, it is a switch. Availability and Flexibility Cisco UCS provides an active-active network fabric out of the box, but it also provides a fully redundant unified management subsystem. Dell s I/O subsystem is compromised by a traditional design that is basically a slight modification of its rack servers. As with other similar designs, the FX2 platform s I/O connections are fixed and physically mapped in the midplane and cannot be modified in any way. For example, Dell network interface card (NIC) port 1 goes to interconnect bay 1, and Dell NIC port 2 goes to interconnect bay 2. Any changes in connectivity between a blade and the outside world forces customers to manually remove or add and recable the physical switches and server I/O adapters. This 2015 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public information. Page 11

12 Architectural Comparison: Cisco UCS and the Dell FX2 Platform Cisco Nexus 5548 Switches Cisco Nexus 5548 Switches Hop 2 Cisco UCS 6248UP Fabric Interconnects Switch ASICs Hop 1 Hop 3 Dell I/O Aggregators Hop 1 A Cisco UCS 2208XP Fabric Extenders X Cisco UCS 5108 Blade Server Chassis One Logical Chassis Up to 160 Servers Dell FX2 Enclosure Two Physical and Logical Chassis 4 Servers Each Figure 7. All Traffic in Cisco UCS Has the Same Low Latency and a Single Network Hop Between Any Two Servers recabling results in application downtime and increased staff time for maintenance tasks. In contrast, Cisco UCS is truly a wire-once technology that allows any changes, including port mapping and I/O assignment, to be fully programmatically implemented and user defined. Using Cisco UCS VICs, Cisco UCS servers can be configured for any workload in minutes, without touching the server. Cisco UCS VICs present up to 256 PCI Express (PCIe) standards-compliant interfaces to the host that can be dynamically configured as either NICs or host bus adapters (HBAs). These adapters can be accessed by the operating system or hypervisor, and they also can be attached directly to virtual machines, accelerating performance. Cisco UCS VICs support fabric failover so that if one fabric fails, the device passes traffic to the alternative fabric, and the operating system never knows about it. You have the option to use OS-based NIC teaming or Cisco UCS hardware fabric failover, or both, depending on application requirements; you can configure this option on a per virtual NIC (vnic) basis. Right-Size Networking Bandwidth for Less Cost Network bandwidth is crucial to getting information to the right place at the right time. Not only do companies need the flexibility to meet current application bandwidth requirements, but they must also be prepared for future bandwidth growth. Rather than a rigid, fixed-i/o topology that requires customers to add more in-chassis switches, Cisco unified fabric allows any server in the chassis to access the total uplink bandwidth to accommodate traffic bursts. For example, you can double the bandwidth simply by increasing the number of uplinks between the chassis and the fabric interconnects virtually and programmatically, with no downtime or application impact. No planning, configuration, or cabling to the chassis switch is required. Cisco UCS can scale blade bandwidth as applications demand: up to 40 Gbps with the mezzanine LOM-format Cisco UCS VIC 1340, or up to 80 Gbps with the Cisco UCS VIC 1380, and more bandwidth is available to full-width (160 Gbps) or full-width double-height (320 Gbps) blade servers Cisco and/or its affiliates. All rights reserved. This document is Cisco Public information. Page 12

13 You are not forced to purchase and overprovision in-chassis switches regardless of the application requirements. Because Cisco UCS architecture requires fewer components to scale your bandwidth for peak traffic flows, you don t have to pay for components or bandwidth that you don t need. Management Comparison Unified management is essential to efficient deployment and ongoing management of blade servers. Because they are meant to be high-density servers, you will be deploying more of them in a smaller footprint. Management Cabling With the Dell solution, each chassis requires a separate management cable from the chassis to the ToR switch, which requires more cabling and switch ports than with Cisco UCS. You can daisy-chain these management cables to reduce the number of upstream ports required, but not the number of cables. Daisy chaining also makes your management network fragile. If one management unit or cable fails, management connectivity to all the downstream units is lost, and you lose connectivity to any device lower on the daisy chain (and cooling fans are set to their maximum speed). With Cisco UCS, management connectivity is carried over the same unified fabric connections made to each chassis, and it is redundant by design and requires no additional cabling. Management Complexity In the Dell solution, each chassis has a chassis management controller (CMC), and each blade has an integrated Dell remote access controller (idrac). For access to advanced features such as power management, Lightweight Directory Access Protocol (LDAP), and two-factor authentication, you need idrac Enterprise, at a cost of US$332 per server per year. At a higher level of management, Dell OpenManage continues to be the standard management option. However, OpenManage does not have flexible policies or templates or support for stateless servers. To obtain advanced features, such as basic configuration templates, you are required to purchase Dell Active System Manager (ASM), with an additional per-server, per-year licensing cost. Dell ASM uses a top-down, scripted approach with limited configurability. ASM does not have a notion of policies, nor does it offer configurable role-based access control (RBAC). Dell has made multiple changes to its management strategy in the past few years, including changes in various high-level tools. Simplified Management With Cisco UCS, servers, connectivity, and management are inseparable. The complete abstraction of configuration information creates an on-demand, zerotouch environment. Cisco UCS was designed from the beginning with embedded, all-inclusive, model-based management through Cisco UCS Manager. Cisco UCS is intelligent infrastructure that is self-aware and self-integrating. Every server connected to Cisco UCS, whether it is a blade server or a rack server, is automatically detected and placed in a resource pool and even automatically configured if you so desire. The system is built from the foundation so that every aspect of server identity, personality, and connectivity is abstracted and can be 2015 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public information. Page 13

14 applied through software using a Cisco UCS service profile. With Cisco UCS, servers are configured automatically, eliminating the manual, time-consuming, errorprone assembly of components into systems. With Cisco VICs, even the number and type of I/O interfaces are programmed dynamically, making every server ready to power any workload at any time. Cisco service profiles can be quickly created from templates, enabling fast configuration of one or 100 blade and rack servers in just a few minutes. Cisco UCS Manager provides integrated, model-based management. With Cisco UCS Manager, administrators manipulate a model of a desired system configuration and associate a model s service profile with hardware resources, and the system configures itself to match the model. This automation accelerates provisioning and workload migration, delivering accurate and rapid scalability. For the first time, you have an automated, policy-based mechanism for aligning the server configuration with the workload. The result is increased IT staff productivity, improved compliance, and reduced risk of failures due to inconsistent configurations. Cisco UCS Manager can be accessed through a GUI, a command-line interface (CLI), or an open, standards-based XML API that is used by a large ecosystem of management tools. Cisco UCS Central Software, an extension of Cisco UCS Manager that uses this XML API, enables management of multiple Cisco UCS deployments across geographic locations, with the first five Cisco UCS Central Software licenses free. Performance and Choice of Processors If you want to standardize on a blade server solution, you should be sure that your vendor supports a variety of blades to enable excellent performance for all your enterprise workloads. Right now, Dell supports three Intel Xeon processor equipped server blocks Dell PowerEdge FC430, FC630, and FC830 all using the Intel Xeon processor E5 v3 family. Smaller servers such as the FC430 cannot support the highest-performing processors, with more than 14 cores. The Dell solution also does not support Intel Xeon processor E7 family CPUs. Alternatively, Cisco UCS offers five different blade servers using a variety of Intel Xeon processor families to meet the size and performance needs of your enterprise workloads. In addition, in just over five years since our first customer shipment, Cisco UCS with versatile Intel Xeon processors had captured more than 100 world performance records, across a wide variety of workloads, with first-to-market results or results that exceed those set by other system vendors, including Dell, HP, and IBM/Lenovo, as of the date of disclosure. Although all these servers use Intel Xeon processors, the architectural advantage of Cisco UCS contributes to the system s world-record-setting performance and the timeliness with which Cisco delivers performance results Cisco and/or its affiliates. All rights reserved. This document is Cisco Public information. Page 14

15 Conclusion The vision for blade servers was a resource pool of servers sharing a common set of environmental, networking, and storage resources, rather than replicating all resources for all servers, as is the case with rack-mount servers. This approach would then reduce both CapEx and OpEx. Dell has tried to implement this vision twice: first with the Dell M1000e chassis and now with the Dell FX2 platform. Traditional blade architectures such as the FX2 originally evolved from the idea of repackaging rack servers and switching as a smaller form factor. As density and performance increased, so did complexity. As a result, these traditional blade solutions have failed to deliver on the promise of blade servers. Cisco UCS is more than a traditional blade solution. It is the only blade system on the market today that delivers true unity computing while also giving you choice. This radically simplified solution is intelligent infrastructure that is automatically configured through integrated, model-based management to accelerate deployment of all your enterprise-class applications and services running in bare-metal, virtualized, and cloud-computing environments. For More Information For more information about Cisco UCS visit Americas Headquarters Cisco Systems, Inc. San Jose, CA Asia Pacific Headquarters Cisco Systems (USA) Pte. Ltd. Singapore Europe Headquarters Cisco Systems International BV Amsterdam, The Netherlands Cisco has more than 200 offices worldwide. Addresses, phone numbers, and fax numbers are listed on the Cisco Website at Cisco and the Cisco logo are trademarks or registered trademarks of Cisco and/or its affiliates in the U.S. and other countries. To view a list of Cisco trademarks, go to this URL: Third party trademarks mentioned are the property of their respective owners. The use of the word partner does not imply a partnership relationship between Cisco and any other company. (1110R) LE /15