Glossary of VMFS terms
For a proper CID chain understanding let's have a look at CID itself. CID stands for Content Identification. It's a 32-bit, randomly created value, that VMware generates every time when it opens the wmx. file. Along with a new snapshot Vmware creates delta.vmdk files and unites them in a chain with the help of CID values. Those values are not constant and are modified every time VM restarts, but hierarchy of parent – child CID Values in the chain must be kept. In the 3 component CID chain with Parent .vmdk, 1st child. Vmdk and 2nd Child .vmdk created after first and second snapshots respectively the submission must be the following:
child vdisk-000002.vmdk points to the child vdisk-000001.vmdk and vdisk-000001.vmdk points to the parent vdisk.vmdk.
If this CID chain is broken VM won't be able to mount virtual disks assigned to it, until the order is reconstructed!
CLI or Command-Line Interface
is a powerful management tool forVMware users to administer VMware ESX/ESXi infrastructures. For those who are used to working from the command line, vSphere Command Line Interface is a strong alternative to such managing tools as vSphere Client. It helps you to run most of the common system administration commands preserving network access to ESX/ESXi systems. vSphere CLI comprises the ESXCLI command set, vicfg- commands, etc., providing more freedom in ESX/ESXi features configuration.
Internet Small Computer System Interface is a transport layer protocol that sends SCSI commands over local-area networks (LANs), wide-area networks (WANs) or the Internet to link data storage devices, ISCSI envelopes commands and bundles the data in packets before sending them over a TCP/IP network, employing a point-to-point connection. After the arrival packets are uncovered, the SCSI commands are separated and the operating system sees the storage as a local SCSI device. So, it allows to unite separate storage into centralized storage array, while hosts perceive it as locally attached disks. iSCSI allows two hosts e.g. web server and database exchanging commands via IP network.
ISCSI doesn't depend on additional expensive equipment and uses standard Ethernet, not requiring specialized skills to install and configure everything properly.
Despite iSCSI great popularity there are more expensive alternatives such as: Fibre Channel over IP (FCIP), Internet Fibre Channel Protocol (iFCP), Fibre Channel over Ethernet (FcoE), ATA over Ethernet (AoE).
stands for a Logical Unit Number. It is a number that identifies separate devices (logical units) addressed by SCSI, iSCSI or Fibre Channel protocol. A LUN can be used with any reading/writing devices, but more often this term is used to refer to the logical disk. Notice that one and the same virtual volume in the RAID array can have different LUN values assigned by different hosts. It's also possible to come across identical LUNs in one host, that belong to the different SCSI Target IDs.
Network File System providing an opportunity for a user on a client computer view, update and store files on remote systems over a network as if they were stored locally on user's own hard drive. The user's system requires NFS client and the other computer needs the NFS server to be mounted. TCP/IP must be also installed on both computers since NFS uses it for files exchange.
More often NFS is employed to configure a central server where users store all their home directories. Further they can be accessed by logging into a client anywhere on the network. NFS also allows to install software products on several computers at once from a centralized source.
There is an extended NFS version - WebNFS which is considered to be more efficient when accessing Web pages and other internet content.
is a copy of the virtual machine disk file that reveals the state and data of the VMDK at a certain point of time. The snapshots are mostly used to provide a detailed information about a virtual disk at a given time if there is a need to restore a virtual machine in the case of any failure or system errors. Though the snapshots must not be treated as providing backups. They are not absolute copies of the original VMDK files. If the base disks become unavailable, the snapshots appear to be useless. Snapshot files can't be bigger than the original disk file, but, nevertheless, demand some extra disk space. This space grows when disk writing activity increases. Snapshots are usually deleted within an hour. Those that stay longer are advised to be erased in 24 hours. 72 hours is a deadline of existence for any snapshot to provide normal VM operation. Though VMware operation is calculated to support up to 32 snapshots united in a chain, it's strongly recommended not to use more than 2 or 3 of them for further backup purposes.
Zettabyte File System is one of the most advanced combinations of a logical volume manager and properly a filesystem. It features high storage capacities: up to 16 Exabyte file size and 256 Zettabyte volume size. ZFS provides powerful data compression, protects data from corruption, verifies data for integrity and performs automatic repair if needed. It also grants such features as snapshots and copy-on-write clones. ZFS consists of the following structural elements:
- vdev – elements making up a pool
- pool – nexus that manifests the underlying storage devices
- dataset – the pool's child filesystem
- volume – the pool's child, functioning as a virtual block device
- snapshot - the state of a dataset or a volume frozen at a point in time
- clone – a new dataset or a volume derived from a snapshot
Being characterized as the last word in filesystems, ZFS is secure, stable and really fast.
Realization or RAID5 architecture on the file system level. Raidz is a software RAID based on ZFS which provides a data\parity distribution scheme similar to RAID 5. Raidz offers some performance improvements compared to RAID 5.
Actually Raidz can act like RAID1, RAID6(raidz2) and RAID7(raidz3), however currently we support only raidz with single parity version.