VMware virtualization-network and storage features

A virtual network can carry multiple protocols and use the corresponding network protocols, fc san, iSCSI SAN, and so on for different applications. The vsphere 5 network consists of the following components: vSphere Standard switch: vsphere Standard switch, which is the default vswitch created by vsphere to manage VM networks vsphere Distributed switch: vSphere Distributed switch, VM and VMKernel networks can be managed, and hosts in the cluster can share and manage a distributed switch, that is, as long as a distributed network is configured on an ESXi host, the cluster automatically applies to other hosts in the cluster. You do not need to configure each ESXi host. Port and port group: port and port group are the logical units provided by the switch to manage VMS and VMkernel. VMKernel Port: The vmkernel port must be configured with IP addresses and gateways. It is mainly used to manage networks, iSCSI storage networks, VMotion, NFS storage, and FT networks, multiple VMKernel networks can be established to separate each network. VM Port Group: VM Port Group enables VM to communicate with VM or VM and physical network communication www.2cto.com. in planning a virtual network, in addition to the existing physical network environment, you may need to consider the following issues: 1. independent Management network required? 2. whether to use an independent network for VMotion 3. whether iSCSI, NFS, or NAS uses an independent network 4. number of physical NICs on the ESXi host, and whether the host can run in a 10 Gb environment. whether to enable the FT function and whether the FT function uses an independent network. 6. whether a physical network has VLANs and whether to integrate VLANs into a virtual environment. A Virtual Switch must contain ports, port groups, and uplinks. If no port or port group exists, it cannot communicate with VMkernel and VM, without UPlinks, the VM cannot communicate with the Internet, so both are required. Port groups are used to divide different types of traffic. uplinks provides communication between virtual machines and physical networks. virtual machines in the same vswitch without uplinks can communicate with each other, however, virtual machines cannot communicate with virtual machines in the physical network or in another vswitch. The Management Network communicates through the VMkernel Port. By default, a vswitch is created when the ESXi host is installed, a VMkernel port is created, and the vmkernel port is set to be used to manage the network. The VMkernel network can be used not only for network management, but also for VMotion, iSCSI, NFS, and FT networks. The NIC binding function provides Nic redundancy, Server Load balancer, and traffic control functions. However, the Server Load balancer is not based on the traffic of each network adapter, but based on the number of connections of the Network Adapter. The supported Server Load balancer policies include vSwitch port-based, Source MAC-based, IP hash-based, explicit failover order. VSwitch Port-based Server Load balancer: the Default Server Load balancer policy. This policy uses the same physical Nic for traffic from a specific vswith port. Source MAC-based load balancing: similar to the vSwitch port-based policy, this policy maps all network connections of the same Source MAC to the same physical Nic. IP hash-based load balancing: implements a hash algorithm for the source IP address and the target IP address, and then maps the traffic to the corresponding physical NIC based on the obtained hash value, that is to say, when the same virtual machine communicates with different destination addresses, different physical NICs may be used due to different hash values. When using this policy, www.2cto.com must note that all uplinks must be connected to the same vswitch, and the switch must be configured with link aggregation. LACP and port aggregation are not supported. Finally, all physical NICs must be set to active, instead of some active standby. Explicit failover order: uses a clear failover sequence. This policy controls traffic in the active, standby, and unused sequence. By default, all virtual machines use the bandwidth of the physical Nic, no restrictions. Traffic control provides average bandwidth, peak bandwidth, and burst size settings to limit the occupation of network bandwidth by virtual machines. Vmware vsphere 5-storage features mainly consider the following aspects about storage performance: storage capacity, storage bandwidth, storage IOPS, storage latency, random read/write IOPS of common disks: 7200 rpm sata about 80 IOPS 10 k rpm sata/SAS/fiber Channel: 120 IOPS 15 k rpm sas/fiber channel: 180 IOPS Based on MLC (multi-layer unit) SSD: about-IOPS Based on SLC (single-layer unit) SSDS: About-IOPS ESXi hosts can mount multiple Storage types: Local SAS/SATA/SCSI hard disks, FC, FCOE, ISCSI, NAS, etc. Of course, the server also needs the corresponding HbA card for support. Www.2cto.com Virtual File System (VMFS) is a virtual file system. It is used to create a unified file system to run virtual machines. VMFS is a cluster file system, which is generally like NTFS, ext3 and other file systems are not cluster file systems. Currently, VMFS 5 supports 64 TB for a single LUN, while VMFS 3 supports only 1 MB for 2 tb vmfs 5, the size of a single file can reach 8 kb for 2 tb vmfs 5 and 64 kb for vmfs 3. VMFS 5 uses a bare rding (RDMs) disk larger than 2 TB, VMFS 3 has a maximum RDMs size of 2 TB. When creating a virtual machine, the format of the virtual hard disk is as follows: streamlined configuration: No matter how large the disk is allocated, the actual storage space occupied is the current storage space, that is, the actual storage space. When the client has input and output, VMkernel first allocates the required space and clears it. That is to say, if you use a streamlined configuration, you need to wait for the allocation space when there is IO, operations can only be performed after the two steps are completed. For applications with frequent I/O operations, the performance will decrease, although saving storage space. After the backup delay is set to zero: if the delay is set to zero, the space is allocated first, but no operation is performed, when I/O exists, I/O can be completed only after the zeroth operation is complete. The latter is set to zero: when the latter is set to zero, the space is allocated and the latter is set to zero. In this way, when there is IO, you do not have to wait for any operation to proceed directly. This article is from Waydee's blog