based on Dell Open Network Ethernet Switch, American ESG Laboratory conducted practical operation test on Big Switch Networks Big Cloud Fabric(BCF), this paper mainly verifies whether BCF's application-centered network design can bring automation, elastic expansion and simplicity to VMware and OpenStack data center environments.
according to the latest ESG survey, 39% of the interviewees pointed out that network management was one of their most important investment in network infrastructure in 2015, other key investments include upgrading the core network of the data center (34%) and network analysis (28%). Server, application, and desktop virtualization are already everywhere, posing increasing challenges to network infrastructure management. In a highly flexible virtualization world, network administrators are difficult to provide appropriate services for applications and users due to the continuous dynamic configuration. As the network expands to accommodate private and public cloud infrastructures, this problem is even worse.
Traditional data center network topology is built based on chassis-type core switches and independent rack-type top-type distribution switches, which may affect all aspects of the data center network (from deployment to continuous management) it brings unnecessary complexity and difficulties. Modern data centers adopt a highly flexible virtualization infrastructure, bringing a variety of network challenges to IT departments. It is quite difficult to use traditional network devices and tools to build and manage modern data center topologies because these tools are designed for a relatively static physical infrastructure.
Solution: Big Cloud Fabric
BigSwitch Cloud Fabric is a specially designed network structure, which supports physical and virtual workloads, allows free selection of coordination software, and provides layer -2 Exchange, layer -3 routing and layer -4-7 service insertion and linking functions. The following figure shows the architecture:
the architecture adopts a branch-backbone topology. The trunk layer is equivalent to the chassis-type switch or the backboard of the router, while the branch layer is equivalent to the combination of the chassis-type router and the line cards in all aggregate switches and edge switches. Each physical branch switch is connected to the trunk switch. The high availability of BCF controller is equivalent to the management card of core router or chassis switch, which is responsible for managing all BCF components. In addition, the BCF controller cluster runs on standard servers and is responsible for the control, management, and policy of switches.
For enterprise-level data centers, BCF adopts the design principle of ultra-large-scale data center network. The main benefits of data centers include centralized management, simplified multi-tenant configuration, SDN based on software and hardware separation, cloud management platform integration, horizontal scaling and elasticity, adaptability, workflow automation and integration, etc.
Big Cloud Fabric uses Dell's Open Network Ethernet switch to provide a comprehensive switch configuration combination from a single supplier, thus simplifying acquisition, installation and support. The height of all Dell open network switches is a rack unit, designed for enterprises with hot plug redundant power supplies and system fans. Table 1 lists the different port combinations of each vSwitch in detail.
in view of the above characteristics, ESG laboratory has carried out practical operation tests on Big Cloud Fabric, and the goal is to verify the agility and simplicity of this new application-centered network structure.
BCF configuration and basic operations
Big Switch Fabric interface provided, the operation is simple and the display is intuitive. As shown in the following figure, the GUI of the controller shows the topology of the network:
test significance: IT managers, senior executives, and relevant business personnel are constantly looking for ways to improve resource efficiency and ROI. Due to the implementation of modern cloud computing and virtualization technologies, enterprises lack professionals with appropriate skills, further upgrading the above requirements to challenges.
Test results: ESG laboratory confirmed that Big Cloud Fabric can transform from bare metal hardware to a fully functional enterprise data center network structure within 15 minutes with only a few clicks. Its GUI and deep analysis are intuitive and easy to use, allowing us to quickly understand the logical and physical network topology, and quickly adjust the system to achieve optimal performance. By aggregating all network hardware into a single management domain, BCF simplifies the IT infrastructure and the workload of personnel.
logical tenant, CIDR block, and router
test significance: network architects face great challenges in implementing logical network design. Manual processing not only consumes a lot of time and energy, but also requires in-depth expertise. Each physical machine needs to be configured separately. If BCF can successfully deploy applications, it will benefit the data center greatly in terms of logical tenants, CIDR blocks, and routers.
Test results: the ESG Lab confirms that BCF can automatically convert the logical abstraction of tenants, logical CIDR blocks, and logical routers into correctly optimized configurations, and apply it to all affected branches and master switches in the network structure. The system automatically applies configuration changes to eliminate the machine-by-machine, time-consuming, and error-prone processing process in the traditional network structure environment.
Programmable and automation
Big Cloud Fabric controller provides standard REST API for system management. All Big Cloud Fabric operations (installation, configuration, and upgrade), whether performed through GUI, CLI, or automated scripts, are sent to the controller through standard REST API, as shown in the following figure:
significance: virtualization and cloud management platforms rely heavily on automation to reduce management workloads and implement a modern self-service IT infrastructure. With the increasing scale and complexity of the network and data center environment, automation becomes more and more critical. However, IT personnel face a challenge-like other parts of the management environment, the management of the network infrastructure can also achieve simplicity and high efficiency.
Test results: BCF is directly integrated with VMware vSphere, NSX, CloudStack, and OpenStack, enabling administrators to extend the network structure from physical structure to virtual structure and simplify configuration and management. Further integration can be achieved through automation: all BCF management functions can be accessed through REST API.
The Chaos Monkey method is used to prove the adaptability of BCF. In the 7-minute test, more than 640 component failures were introduced in total, and the performance was not affected. With redundancy, the entire network structure software can be upgraded with one command without interrupting normal operation. By integrating adaptability, automation, and integration into the cloud management platform, BCF reduces the burden on IT infrastructure and employees.
Organizations face challenges to build networks that respond to business needs of highly virtualized data centers, public clouds, and private clouds. In this case, the Big Switch Networks provided by Big Cloud Fabric, combined with Dell open network switches, can provide a simple, scalable, elastic and highly available network, it is not only a feasible alternative for traditional chassis core switches and traditional data center topologies, but may even be better than traditional solutions.
The above briefly introduces the test results of EGS on BCF configuration and basic operation, application deployment, programmable and automation. For more detailed test contents, methods and results, please refer to the PDF document.
As one of the top ten research and analysis consulting companies in the world, ESG has been providing technical manufacturers, IT professionals and institutional investors with strategic guidance on data center technology and business development. In addition, it also provides professional research for all links of the whole industry chain to ensure that users make the most correct choice.
This article is forwarded from d1net (reprinted)