Saturday, November 27, 2010

Data Centre Concepts - Servers, Storage Networks, Networking and Virtualisation

With my background primarily in applications development I've been educating myself with terminology used in enterprise data centres over the past few months in order to demystify the language used by my IT Pro colleagues:

  • Blade - stripped down server to minimise space and energy usage
  • Enclosure/Chassis - provides power, cooling, networking and management
  • Rack Unit - 1U = 48cm x 4.45cm
  • Half-Height Blade - 1/2 size of traditional blades
  • p-Class Blade - HP ProLiant blade servers
  • c-Class Blade - HP's next generation blades
  • iLO (Integrated Lights-Out) - HP's out-of-band server management technology used to perform activities on a HP server e.g. to reset the server if the server doesn't respond anymore via the normal network card or power-up the server from a remote location even if the server is shut down. Physically connected via an Ethernet port on the iLO card
  • IBM CEC (Central Electronics Component) p570 - a module housing CPU's. RAM, PCI backplane etc. linked by scalability cables.

Storage Networks

  • SAN (Storage Area Network) - storage device accessible to servers so the devices appear as locally attached to the O/S
  • NAS (Network Attached Storage) - use NFS/SMB where it is clear that the storage is remote and computers request a portion of an abstract file rather than a disk block
  • DAS (Direct Attached Storage) - directly attached to a server or workstation, without a storage network in between
  • LUN (Logical Unit Number) - a logical disk as created on a SAN
  • NFS (Network File System) - a protocol primarily used in Unix/Linux allowing a user on a client computer to access files over a network in a manner similar to how local storage is accessed
  • SMB/CIFS (Server Message Block / Common Internet File System) - a protocol primarily used in Windows to provide shared access to files, printers, serial ports, and miscellaneous communications between nodes on a network
  • FC (Fibre Channel) - a gigabit-speed network technology that has become the standard connection type for SAN
  • Switched Fabric - nodes connected to each other via one or more network switches offering better total throughput than broadcast networks because traffic is spread across multiple physical links
  • Fibre Channel Switch - a network switch compatible with the FC protocol allowing for the creation of a Fibre Channel fabric
  • FC-SW (Fibre Channel Switched Fabric) - devices connected to each other through one or more Fibre Channel switches. Usually in a mesh network with devices on the edges/leaves of the mesh. Two separate fabrics can be employed to provide failover capability for redundancy
  • Fibre Channel Zoning - the partitioning of a FC fabric into smaller subsets to restrict interference, add security, and to simplify management. While a SAN makes available several LUNs, each system connected to the SAN should only be allowed access to a controlled subset of the LUNs
  • RSCN (Registered State Change Notification) - a FC fabric's notification sent to all specified nodes in case of any major fabric changes. This allows nodes to immediately gain knowledge about the fabric and react accordingly
  • LUN Masking - an authorization process that makes a LUN available to some hosts and unavailable to other hosts
  • FCP (Fibre Channel Protocol) - a transport protocol which predominantly transports SCSI commands over FC networks
  • iSCSI (SCSI over TCP/IP) - an IP based storage networking standard for linking data storage facilities. Unlike traditional FC, which requires special-purpose cabling, iSCSI can be run over long distances using existing network infrastructure
  • SPI (Parallel SCSI) - a parallel SCSI bus technology
  • SAS (Serial Attached SCSI) - a point-to-point serial protocol that replaces SPI
  • PATA (Parallel Advanced Technology Attachment) - bus interface for connecting adapters to storage devices
  • SATA (Serial ATA) - replaced PATA with serial cable reducing cable bulk with faster data transfer and hot-swapping. eSATA provides external connectivity
  • FATA - low cost SATA disk equipped with converter that changes the interface to FC to allow it to be used in an enterprise class disk enclosure
  • FC Disk - native FC disk providing greater IOPS (FC 15,000 rpm drive is likely to deliver IOPS around 4.5 times greater than a FATA 7,200 drive)
  • IOPS (Input/Output Operations Per Second) - common benchmark for hard disks. Comparative measurements include Total IOPS, Random Read IOPS, Random Write IOPS, Sequential Read IOPS, Sequential Write IOPS
  • Disk Latency - the delay between receiving the instruction and retrieving the data. Determined by a sum of the spin-up time, seek time, rotational delay, and transfer time
  • RAID (Redundant Array of Independent Disks) - provides increased storage functions and reliability through redundancy achieved by combining multiple disks into a logical unit, where data is distributed across the drives in one of several ways called "RAID levels"
  • EVA (Enterprise Virtual Array) - HP's virtualised storage architecture offering
  • FCoE (FC over Ethernet) - an encapsulation of FC frames over Ethernet allowing FC to use 10 Gigabit Ethernet (or higher speeds)
  • HBA (Host Bus Adapter) - connects a host system to other network and storage devices. Each HBA has a unique WWN
  • SRM (Storage Resource Management) - process of optimising the efficiency and speed with which the available drive space is utilised in a SAN. SRM identifies underutilised capacity, identifies old or non-critical data that could be moved to less-expensive storage, and helps predict future capacity requirements
  • Ethernet - a family of frame-based networking technologies for LANs. Defines wiring and signaling standards for the OSI Physical Layer and a common addressing format and Media Access Control at the Data Link Layer
  • Subnet - a logically visible subdivision of an IP network
  • VLAN (Virtual LAN) - a group of hosts that communicate as if they were attached to the same broadcast domain regardless of their physical location
  • Ring Topology - a topology in which each node connects to exactly two other nodes forming a single continuous pathway
  • Mesh Network - a topology in which each node may act as an independent router allowing for continuous connections and reconfiguration around broken or blocked paths
  • Upstream Server - a server that provides service to another server
  • STP (Spanning Tree Protocol) - ensures a loop-free topology for any bridged Ethernet LAN allowing a network to include redundant links to provide automatic backup paths if an active link fails, without the danger of bridge loops, or the need for manual enabling/disabling
  • Trunking / Link Aggregation - Cisco uses the term "trunking" to mean "VLAN multiplexing" - carrying multiple VLANs through a single network link through the use of a "trunking protocol". To allow for multiple VLANs on one link, frames from individual VLANs must be identified
  • DTP (Dynamic Trunking Protocol) - Cisco proprietary protocol to negotiate trunking on a link between two VLAN-aware switches and the type of trunking encapsulation to be used
  • VLAN Trunking Protocol (VTP) - Cisco proprietary protocol to manage VLAN crud changes across network. VLAN changes on VTP server distributed through all switches in domain. Equivalent to IEEE GVRP/MVRP.
  • TCP Wrappers - a host-based Networking ACL system used to filter network access to Internet Protocol servers on Unix-like operating systems
  • Catalyst 6500 Switch - a modular network switch manufactured by Cisco comprising of a chassis, power supplies, one or two supervisors, line cards and service modules
  • WWN (World Wide Name) - a unique 8 byte number which identifies a particular FC, ATA or SAS target
  • MAC (Media Access Control) Address - a unique identifier assigned to network interfaces for communications on the physical network segment
  • ARP (Address Resolution Protocol) - protocol frequently used to translate IPv4 addresses into MAC addresses
  • POP (Point of Presence) -a physical access point to the Internet that houses servers, routers, ATM switches and digital/analog call aggregators
  • SNMP (Simple Network Management Protocol) - a protocol for monitoring and managing routers, switches, servers etc. on IP networks by exposing variables describing the system configuration
  • VMWare ESX / Hyper-V - enterprise-level virtualisation hypervisors
  • VHD / VMDK - virtual hard disk file formats
  • AMD-V / VT-x - x86 processor extensions offering hardware support for virtualisation
  • Memory Ballooning - a process by which the host of a VM influences the memory management of the VM's operating system
  • VMWare VMotion - the capability to move a running VM from one ESX host to another
  • VMWare DRS (Dynamic Resource Scheduler) - automatic load balancing of a ESX cluster using VMotion
  • VMWare HA (High Availability) - in case of hardware failure in a cluster VM's automatically restart on another host in the cluster
  • VMWare VSphere - cloud operating system able to manage large pools of virtualized infrastructure

No comments: