Argentina
Australia
Austria
Belgium
Brazil
Bulgaria
Canada
Chile
Colombia
Cyprus
Czechia
Denmark
Ecuador
Egypt
Estonia
Finland
France
Germany
Great Britain
Greece
Hungary
India
Indonesia
Ireland
Italy
Japan
Latvia
Lithuania
Luxembourg
Malaysia
Malta
Mexico
Netherlands
New Zealand
Norway
Philippines
Poland
Portugal
Romania
Russia
Singapore
Slovakia
Slovenia
South Africa
South Korea
Spain
Sweden
Switzerland
Taiwan
Thailand
Turkey
Ukraine
United States
Hersey Cartwright
kim bottu
In this chapter, we will cover the following topics:
Determining network bandwidth requirements
Standard or distributed virtual switches
Providing network availability
Network resource management
Using private VLANs
IP storage network design considerations
Enabling jumbo frames
Designing for VMkernel services
Creating custom TCP/IP stacks
vMotion network design considerations
IPv6 in a vSphere design
In order to effectively design a virtual network infrastructure, a design architect must understand the virtual network architecture, including which features are available and how they are configured. This chapter contains recipes that a design architect can use to design a virtual network architecture that provides the capacity and availability required to support the virtual infrastructure.
The logical network design includes the calculation of the network capacity or the bandwidth required to support the virtual machines and the determination of the capacity required to support VMware technologies, such as vMotion and fault tolerance. If IP-based storage connectivity is required, the design must account for the networking required to support the storage traffic as well. The following diagram displays how a network design is integrated into the design process:
This chapter discusses the different virtual network switch technologies available in vSphere and the different features...
Bandwidth refers to the capacity of the network and is measured in either gigabits per second (Gbps) or megabits per second (Mbps). The bandwidth required is based on the amount of data transferred or the throughput required by the virtual machines. Most modern networks are capable of transferring data at 1 Gbps or 10 Gbps. Network adapters that support 40 Gbps have recently become available.
The number of physical network adapters in each host required to support a solution is dependent on the amount of bandwidth required to support virtual machine network traffic, the number of virtual switches required, and the network redundancy requirements.
From the case example in Chapter 3, The Design Factors, the following information is used to help calculate the network bandwidth requirements:
Cisco switches are used for network connectivity. Separate VLANs exist for management connectivity and production application connectivity.
No more than 20 application...
The connectivity of the virtual network to the physical network in a vSphere environment is accomplished using one of two virtual switch technologies: the standard virtual switch (vSwitch) or the virtual distributed switch (vDSwitch). VMware technologies such as VMware HA, VMware DRS, and fault tolerance require that virtual switch configurations be consistent across all ESXi hosts in a cluster.
Identify the features and capabilities of virtual standard switches and distributed virtual switches.
Based on the design requirements, determine which virtual switch technology should be selected to support them.
The virtual switch technology chosen is dependent on the connectivity, availability, manageability requirements, and the features available in the virtual switch.
A standard virtual switch is configured and managed independently on each ESXi host and supports up to 1024 virtual switch ports per vSwitch. Because vSwitches are...
Network availability is obtained by minimizing Single Points of Failure (SPOF) and providing sufficient capacity. Multiple network ports, network adapters, and physical switches can be used to minimize single points of failure, and link aggregation can be used to provide load balancing across multiple network adapters.
vSphere virtual network configurations offer multiple NIC teaming and load balancing options. The options used are dependent on the number of network adapters available, the number of virtual machines connected, the physical network's topology, and the amount of bandwidth required.
Identify the availability options on virtual switches and virtual switch port groups.
Determine the load balancing policies to provide availability based on the design requirements.
Determine the network adapter teaming policies to provide availability based on the design requirements.
In a vSphere environment, physical network resources are shared across multiple virtual machines and services. The ability to ensure that sufficient capacity is available across shared resources, therefore, becomes important. If a single virtual machine or a VMkernel network service, such as vMotion or fault tolerance, saturates the available network capacity, other virtual machines and services, including host management services, may be adversely impacted.
Identify the traffic shaping and network resource controls available in the virtual network switches.
Determine the network resources required for different traffic types: management, IP storage, vMotion, and virtual machine traffic.
Design traffic shaping, Network I/O Control (NIOC) policies, and Network Resource Pools to guarantee or limit network resources for the network traffic types based on the design requirements.
Private VLANs (PVLANs) are an extension of the VLAN standard. PVLANs can be configured on virtual distributed switches in order to isolate traffic between virtual machines in the same VLAN.
Identify the types of PVLANs available and the functionalities of each.
Determine the use cases for the PVLANs and identify whether the PVLANs can be used to satisfy the design requirements.
Design the PVLANs to meet the design requirements.
A primary PVLAN is created on a vDSwitch, and secondary PVLANs are associated with the primary PVLAN. There are three types of secondary PVLAN: Promiscuous, Community, and Isolated. They are depicted in the following diagram:
The virtual machine connections in a Promiscuous PVLAN can communicate with all the virtual machine connections in the same primary PVLAN. When a primary PVLAN is created, a Promiscuous PVLAN is created with the same PVLAN ID as the primary PVLAN.
Virtual machine connections in a Community PVLAN can communicate...
iSCSI, NFS, and Fiber Channel over Ethernet (FCoE) are IP-based storage protocols supported in a vSphere environment. This recipe covers the design considerations when designing the IP networks that will be used for storage traffic.
Identify the network connectivity and virtual switch configurations required for IP-connected storage.
Determine the best practices to be used to provide connectivity for IP-connected storage.
Design the IP storage connectivity to meet the design requirements.
IP storage traffic should be separated from other IP traffic. This separation can be provided by either using physically separate hardware (network adapters and physical switches), or separate VLANs for IP storage traffic. The networks associated with IP storage should be directly connected and non-routable.
Multiple network paths to storage should be configured to provide redundancy and load balancing. Single points of failure should be minimized...
Enabling jumbo frames on the networks used for vMotion or IP Storage can increase performance and throughput. When jumbo frames are configured, iSCSI or NFS packets can be transferred over the network in a single frame; there is no fragmentation. This decreases the amount of CPU overhead required to encapsulate and de-encapsulate IP storage packets.
Configure jumbo frames on virtual switches.
Configure jumbo frames on VMkernel ports.
Ensure jumbo frames are configured end to end on the physical network: physical switches and array network interfaces.
Test the network for proper end-to-end jumbo frames configuration.
Jumbo frames must be supported and enabled on the network from end to end; this includes the physical network infrastructure as well. In vSphere, jumbo frames are enabled either in the vSwitch configuration or on the vDSwitch uplinks by setting the Maximum Transmission Unit (MTU) value to 9000...
TCP/IP stacks provide flexibility in the VMkernel interface design by allowing you to apply specific DNS and default gateway configurations to a VMkernel interface on a host.
There are three preconfigured TCP/IP stacks:
Custom TCP/IP stacks can be used to handle the network traffic of other applications and services, which may require separate DNS and default gateway configurations.
Create a custom TCP/IP stack on an ESXi host.
Configure DNS, Default Gateway, and Advanced settings on TCP/IP stack.
Assign TCP/IP stack to a VMkernel adapter.
VMkernel interfaces are configured to provide network connectivity for services in the vSphere environment. VMkernels provide network paths for service connectivity. Multiple VMkernel interfaces can be created to provide physical or logical separation for these services.
Identify services that require a VMkernel interface.
Create a VMkernel interface to support the service.
Enable services on the VMkernel interface.
Most vSphere services require a VMkernel interface to provide network connectivity. These services include the following:
ESXi management
vMotion
Fault tolerance
Virtual SAN
vSphere replication
IP storage (NFS, iSCSI, FCoE)
Multiple services can share a single VMkernel port, or the services can be separated across multiple VMkernel ports for performance, management, and security. Services can be enabled on VMkernel interfaces at the time of creation or by editing Port properties, as shown in the following screenshot:
vMotion allows the running state of a virtual machine to be transferred from one ESXi host to another. The network traffic required for the migration uses the VMkernel interfaces that have been enabled for vMotion. vMotion connectivity between ESXi hosts is required when using Distributed Resource Scheduler (DRS) in order to balance the virtual machine load across hosts in a DRS-enabled cluster.
Determine the best practices to configure the network connectivity required to support vMotion.
Identify the benefits of keeping virtual machines together on the same host in order to minimize the network traffic that must transverse the physical uplinks.
Design the vMotion network connectivity to support the design requirements.
Design DRS rules to support the design requirements.
Internet Protocol version 6 (IPv6) was developed to replace IP version 4 (IPv4). IPv6 addresses are 128-bit IP addresses compared to the 32-bit addresses in IPv4. IPv6 is becoming more common in datacenter network environments, and vSphere has included support for IPv6 since vSphere 5.x.
Enable IPv6 on the ESXi host.
Determine the vSphere features and services with IPv6 support.
Configure VMkernel interfaces to use IPv6.
By default, IPv6 support is enabled on ESXi hosts. If the IPv6 support is changed, disabled, or enabled, a host reboot is required. Enabling or disabling IPv6 is done on each ESXi host by editing Advanced Network Settings from the Networking management tab for the host, as shown in the following screenshot:
Once enabled, IPv6 can be configured for supported vSphere features and services. The following vSphere features and services support IPv6:
ESXi and vCenter Management
vMotion and vSphere DRS
Fault tolerance
vSphere HA
NFS v3 storage...
The rest of the chapter is locked
© 2016 Packt Publishing Limited All Rights Reserved
