Using High Availability Technologies
Lesson 12
Skills Matrix
Technology Skill Objective Domain Objective #
Using Virtualization Configure Windows Server Hyper-V and virtual
machines
1.3
What Is High Availability? Configure high availability 1.4
Creating a Terminal Server Farm
Configure Terminal
Services load balancing
2.3
High Availability
• Application servers often provide network users with tools they need to perform their jobs.
• If a vital server fails, productivity suffers.
• For that reason, many organizations implement high availability technologies that ensure the
continued performance of their server applications.
• High availability typically takes the form of redundant hardware, software, or data
components, which enable an application to
continue running despite disasters such as a drive failure, a power outage, or data loss.
RAID
• RAID (Redundant Array of Independent Disks) is one of the most basic and commonly used high availability technologies.
• A RAID array consists of multiple hard disks, with a controller that enables the array to automatically store redundant copies of data on different drives.
• RAID implementations come in many forms, ranging from workstation solutions, to server controllers, to stand-alone disk arrays that can connect directly to a computer or a storage area network.
Other Redundant Components
• Other high availability components often found in servers include redundant power supplies, active cooling systems, simultaneous connections to
multiple networks, and hot-swappable components that you can replace without shutting down the
computer.
• These are all relatively low-cost solutions that can enable a server to continue running despite a
component failure or a service outage.
• The remainder of this lesson discusses some of the more elaborate high availability technologies that you can use to provide even greater amounts of fault tolerance.
Storage Area Networks (SAN)
• Storage area networks are typically high-end solutions, which enterprise networks use to deploy large amounts of storage and make this storage available to other
connected devices.
• At the highest level, a storage area network (SAN) is simply a network dedicated solely to high-speed
connections between servers and storage devices.
• Instead of installing disk drives into servers or connecting them using a SCSI bus, a SAN consists of one or more
drive arrays equipped with network interface adapters, which you connect to your servers using standard twisted pair or fiber optic network cables.
• A SAN-connected server, therefore, has a minimum of two network adapters, one for the standard LAN
connection, and one for the SAN.
Storage Area Networks (SAN)
Multiple Servers Connected to a SAN
Fibre Channel
• Fibre Channel is a high-speed serial networking technology, originally designed for use with
supercomputers but now associated primarily with storage area networking.
• Fibre Channel is a versatile technology, supporting various network media, transmission speeds,
topologies, and upper level protocols.
• Its primary disadvantage is that it requires specialized hardware that can be extremely expensive.
Fibre Channel Media
• A Fibre Channel network can use a variety of network media.
• Copper alternatives include video or miniature coaxial cable and, more commonly, shielded
twisted pair (STP) with DB-9 or HSSDC (High Speed Serial Data Connection) cable connectors.
• Fiber optic alternatives include 62.5- or 50-ohm multimode and 7- or 9-ohm singlemode, all using LC or SC connectors.
• Because Fibre Channel uses serial instead of parallel signaling, it can span much longer
distances than a pure SCSI connection, up to 50 kilometers or more in some cases.
Fiber Channel Speeds
• Transmission speeds for Fibre Channel
networks range from 133 Mbps (megabits per second) to 1 Gbps (gigabit per second) for copper cables, and up to 10 Gbps for fiber optic.
• Maximum speeds depend on the type of
cable the network uses, the lengths of the
cable segments, and, in the case of fiber
optic, the type of laser used to transmit the
signals.
Fibre Channel Topologies
• Point-to-point (FC-P2P) — Consists of two
devices only, directly connected with a single cable.
• Arbitrated loop (FC-AL) — Consists of up to 127 devices, connected in a loop topology, similar to that of a token ring network.
– The loop can be physical, with each device connected to the next device, or virtual, with each device connected to a hub that
implements the loop.
Fibre Channel Topologies
• Switched fabric (FC-SW) — Consists of up to 16,777,216 (224) devices, each of which is connected to a Fibre Channel switch.
• Unlike Ethernet switches, Fibre Channel
switches provide redundant paths between the connected devices, forming a topology called a mesh or fabric.
• If a switch or a connection between switches
fails, data can find an alternate path through
the fabric to its destination.
iSCSI
• iSCSI is an alternative storage area
networking technology that enables servers and storage devices to exchange SCSI traffic using a standard IP network instead of a
dedicated Fibre Channel network.
• Because iSCSI uses a standard IP network for its lower layer functionality, you can use the same cables, network adapters,
switches, and routers for a SAN as you would for a LAN or wide area network (WAN),
without any modifications.
iSCSI
• iSCSI communication is based on two elements:
– Initiators – Targets
iSCSI Initiator
• Initiates the SCSI communication process.
• Is a hardware or software device running on a computer that accesses the storage devices on the SAN.
• On an iSCSI network, the initiator takes the place of the host adapter that traditional SCSI implementations use to connect storage devices to a computer.
• The initiator receives I/O requests from the operating system and sends them, in the form of SCSI commands, to specific storage devices on the SAN.
• The only difference between an iSCSI initiator and a SCSI host adapter is that the initiator packages the SCSI traffic in TCP/IP packets, instead of using the native SCSI
protocols.
iSCSI Target
• Integrated into a drive array or computer.
• The target receives SCSI commands from the initiator and passes them to a storage device, which is represented by a logical unit number (LUN).
• A LUN is essentially an address that SCSI devices use to identify a specific storage resource.
• A single LUN can represent an entire hard disk, part of a disk, or a slice of a RAID array.
• Therefore, a single computer or drive array can
have many LUNs, represented by multiple targets.
iSNSs
• After the initiators and targets are in place, the only problem remaining in iSCSI communications is how the two locate each other.
• The Internet Storage Name Service (iSNS) makes this possible by registering the presence of
initiators and targets on a SAN and responding to queries from iSNS clients.
• Windows Server 2008 includes an iSNS
implementation as a feature, which can provide the identification service for an entire SAN.
iSNS Components
• iSNS server — Receives and processes registration requests and queries from clients on the SAN, using the iSNS database as an information store.
• iSNS database — Information stored on an iSNS server that contains data supplied by client registrations. The server retrieves the data to respond to client queries.
• iSNS clients — Component in iSCSI initiators and
targets that registers information about itself with an iSNS server and sends queries to the server for
information about other clients.
• iSNS Protocol (iSNSP) — Protocol used for all
registration and query traffic between iSNS servers and clients.
Windows Server 2008 with SANs
• Windows Server 2008 includes several
components that enable the computer to interact with devices on a SAN, as follows:
– iSCSI Initiator — Establishes connections with iSCSI targets on the SAN.
– Internet Storage Name Server — Registers and locates iSCSI initiators and targets.
– Storage Manager for SANs — Enables administrators to manage storage devices on Fibre Channel or
iSCSI SANs that are compatible with the Virtual Disk Service (VDS) using a standardized interface.
– Storage Explorer — Enables administrators to view and manage the devices on Fibre Channel or iSCSI SANs.
Installing Internet Storage Name Server
iSCSI Initiator
Add iSNS Server and Add Target Portal Dialog Boxes
Target Tab of iSCSI Initiator Properties Sheet
Advanced Settings Dialog Box
Storage Manager for SANs Console
Creating LUNS using Storage Manager for SANs
Displaying Subsystem Information
Displaying Drive Information
Storage Explorer
Configuring iSCSI Initiator in Storage Explorer
Clustering Servers
• Server clustering can perform two services on an enterprise network.
• In addition to providing fault tolerance in the event of a server failure, it can provide
network load balancing for busy
applications.
Clustering Servers
• The ultimate in fault tolerance, however, is to have entire servers that are redundant, so that if
anything goes wrong with one computer, another one can take its place almost immediately.
• In Windows Server 2008, this is known as a failover cluster.
• When a Web server or other application becomes overwhelmed by a large volume of users, you can deploy multiple identical servers, also known as a server farm, and distribute the user traffic evenly among the computers using Network Load
Balancing clusters.
Failover Cluster Requirements
• Duplicate servers
• Shared storage
• Redundant network connections
Validating a Failover Cluster Configuration
Validate a Configuration Wizard
Select Servers or a Cluster Page
Testing Options Page
Confirmation page
Summary Page
Failover Cluster Validation Report
Create Cluster Wizard
Select Servers Page
Access Point for Administering the Cluster Page
Confirmation Page
Newly Created Cluster in the Failover Clusters Management Console
Cluster Failover
• After you create the cluster, you can use the Failover Cluster Management console to
specify the applications the cluster will manage.
• If a server fails, the applications you select
are immediately executed on another server
to keep them available to clients at all times.
Network Load Balancing (NLB)
• Network load balancing (NLB) differs from failover clustering because its primary function is not fault tolerance, but rather more efficient support of
heavy user traffic.
• In network load balancing, all of the servers in the cluster are operational and service clients.
• The NLB cluster is a logical entity with its own name and IP address.
• Clients connect to the cluster, rather than the individual servers.
• The cluster distributes the incoming requests evenly among its component servers.
Load Balancing Terminal Servers
• Windows Server 2008 also supports the use of network load balancing for terminal
servers.
• For any organization with more than a few Terminal Services clients, multiple terminal servers are required.
• Network load balancing can ensure that the
client sessions are distributed evenly among
the servers.
Deploying Terminal Services with NLB
• The process of deploying Terminal Services with network load balancing consists of two parts:
– Creating a terminal server farm.
– Creating a network load balancing cluster.
Terminal Services Configuration Console
TS Session Broker Tab
Completed TS Session Broker
Group Policy Settings for TS Session Broker
DNS Round-Robin
• Under normal circumstances, the DNS server always resolves a given name into the same IP address , thereby causing all terminal services clients to connect initially to the same terminal server.
• To prevent performance problems, Microsoft recommends the use of a secondary load
balancing mechanism to distribute the initial
connection attempts among the various terminal servers.
• The most common way to do this is to use the DNS round-robin technique.
Virtualization
• Virtualization enables administrators to deploy server roles on separate virtual
machines that run on a single computer.
• This enables each role to operate within its
own protected environment.
Virtualization
• The process of deploying and maintaining multiple instances of an operating system, called virtual
machines (VMs), on a single computer.
• Each virtual machine contains a completely separate copy of the operating system with its own virtual
hardware resources, device drivers, and applications.
• To the network, each virtual machine looks like a
separate computer with its own name and IP address.
• As a result, you are not combining the security risks of multiple roles in a single operating system instance.
• You update each instance of the operating system separately.
Virtualization Architectures
• Virtualization products can use several
different architectures that enable them to share a computer’s hardware resources
among several virtual machines.
VMM with Hypervisor
Hyper-V
• While Microsoft has designed Hyper-V to be a role included with the Windows Server
2008 operating system, Hyper-V is not
included in the initial Windows Server 2008 release.
• Instead, Microsoft provides it as a separate
download that adds the Hyper-V role to the
operating system.
Hyper-V
• Hyper-V is a Windows Server 2008 role like any other, which you can install using the Server Manager console.
• Hyper-V has hardware and licensing
requirements that go beyond those for the Windows Server 2008 operating system.
• In practice, the technology will largely be limited to enterprise deployments that are willing to make a substantial hardware
investment in virtualization technology.
Hyper-V Requirements
• Hyper-V is included in the Windows Server 2008
Standard, Enterprise, and Datacenter products, but only in the 64-bit versions, for computers with x64 processors.
• There will be no Hyper-V support for computers with 32-bit x86 processors.
• In addition, the hypervisor requires a processor with hardware support for virtualization, which limits the use of Hyper-V to computers with
processors that have a virtualization extension, as well as chipset and BIOS support for virtualization.
Intel has named their virtualization extension VT, while AMD calls theirs AMD-V.
Hyper-V Licensing
• In addition to the specialized hardware requirements for Hyper-V, Microsoft has added a licensing requirement.
• For licensing purposes, Microsoft refers to each virtual machine that you create on a Hyper-V server as a virtual instance.
• Each Windows Server 2008 version includes
a set number of virtual instances; you must
purchase licenses to create additional ones.
Hyper-V Manager Console
Settings Configuring Interface
Summary
• High availability typically takes the form of redundant hardware, software, or data
components that enable an application to continue running even if a disaster occurs.
• A storage area network (SAN) is a network
dedicated solely to high-speed connections
between servers and storage devices.
Summary
• Fibre Channel is a high-speed serial
networking technology that was originally designed for use with supercomputers, but which is now associated primarily with
storage area networking.
• iSCSI is an alternative storage area
networking technology that enables servers and storage devices to exchange SCSI traffic using a standard IP network instead of a
dedicated Fibre Channel network.
Summary
• An iSCSI initiator is a hardware or software device running on a computer that accesses the storage devices on the SAN.
• The other half of the iSCSI equation is the iSCSI target that receives SCSI commands from the initiator and passes them to a
storage device, represented by a logical unit
number (LUN).
Summary
• The Internet Storage Name Service (iSNS) registers the presence of initiators and
targets on a SAN and responds to queries
from iSNS clients.
Summary
• Storage Manager for SANs can manage only storage devices that include support for the Microsoft Virtual Disk Service.
• The storage device manufacturer must
supply a software component called a VDS
hardware provider, which you install on the
computer that will manage the device.
Summary
• Storage Explorer is an MMC console that provides information about SAN resources and enables
administrators to perform a variety of management tasks.
• The ultimate in fault tolerance is to have entire
servers that are redundant so that if anything goes wrong with one computer, another one can take its place almost immediately.
– In Windows Server 2008, this is known as a failover cluster.