120 likes | 242 Views
CHAPTER 7. CLUSTERING SERVERS. CLUSTERING TYPES. There are 2 types of clustering ; Server clusters Network Load Balancing (NLB) The difference between the two types is based on the type of applications the servers must run & the nature of the data they use.
E N D
CHAPTER 7 CLUSTERING SERVERS
CLUSTERING TYPES • There are 2 types of clustering ; • Server clusters • Network Load Balancing (NLB) • The difference between the two types is based on the type of applications the servers must run & the nature of the data they use.
Server clusters (statefull applications) • Designed for application that have long running in memory states or large frequently changing data sets. • Includes database servers such as SQL server, email & massaging servers; microsoft exchange • In server cluster all the computers(nodes)are connected to a common data set such as shared SCSI bus / storage area network. • Since all nodes have access to same application, any of them can process a request from client • Active nodes receives & processes requests from clients while passive stays idle as fallback(instantaneously) should an active node fail.
Network Load Balancing (Stateless application) • Provides high availability & reliability with the addition of high scalability as well. • Intended for applications for small data sets that rarely change & do not have long running in memory states. • Include Web, file transfer protocol, and VPN servers. • Every client request is a separate transaction so its possible to distribute the requests among multiple servers to balance the processing loads. • NBL cluster all have identical cloned data sets and are all active nodes, should one or more fail, the others will take up the tasks. • Scalability is due to the possibility to add more servers to the cluster.
Designing a clustering solution • Access how much availability, reliability & scalability you need. • Why we need clustering solution; • Software failures – application malfunction, interference with other applications, failure applying upgrades, conflict with newly installed program, virus or malicious code.\ • Hardware failures – harddrives, cooling fan, power supplies & other components have limited life span • Site failure – geographically dispersed cluster, with servers are in different building or cities.
Estimating availability requirements • Degree of availability required depends on variety of factors; • Nature of application running • Size • Location • Distribution of user base • Role of application in your organization. • Ex; if the server is said 99% available, that means the application would be unavailable for up to 87.6 hours during a year! • 99.9% = 8.76 hours a year • To achieve high availability, apart from clustering solution, you may have to install fault tolerant hardware, create hardware testing plan & establish operational policies.
Scaling clusters & number of clusters • Both server clusters and NLB are scalable clustering solutions = you can improve the performance of the cluster as the need of your organization grow. • 2 basic methods are; • Scaling up – improving individual server performance by modifying the computers hardware configuration. • Scaling out – adding servers to an existing cluster. • Servers in a cluster can run multiple applications, so; • You may combine multiple applications in a single cluster • Create a separate cluster for each application.
Understanding nbl • Nbl cluster consists up to 32 servers reffered to as hosts. • Each hosts runs a duplicate copy of the application you want the cluster to provide to clients. • NBL works by creating on each host a virtual network adapter that represents the cluster as a single entity.
Nbl network design Internet Internet router Internet router Switch Firewall Firewall Node Node Node Node
Planning a server cluster hardware configuration • For maximum availability having two network interface adapters in each computer is preferable; • One providing connection to client network • One connecting to a network dedicated to communications between the servers in the cluster. • Each server must have a separate connection to the shared storage device. 2 types of storage connection; • Using SCSI – bus architecture used to connect storage devices & other peripherals to personal computers. • Using Fiber Channel – high speed serial networking technology that was originally conceived as general purpose networking solution.
Selecting a quorum model • Every node in a server maintains a copy of the cluster database in its registry. • When a cluster goes offline, the database is no longer updated & the status changes. When the node comes back online, its obtains the current copy of the database to rejoin the cluster, & gets the copy from the cluster’s quorum resources. • Selecting the location for the quorum is a crucial part of creating cluster. 3 quorum models are; • Single-node cluster – a cluster that consists of only 1 server. • Single-quorum device cluster – the cluster uses a single quorum resource • Majority node set cluster – separate copy of quorum is stored in each cluster node.
Configuring failover policies • You may implement a number of different failover policies that control which application nodes an application uses & when. • Failover pairs • Hot standby server • N+I • Failover ring • Random