As more and more application architectures are decoupled and distributed, the requirements and dependencies of load balancer devices are increasing while the component scale-out capability increases, and the load-balancing devices are required to have strong scale-out capability. Load balancing clustering technology is becoming more and more important. While many vendors offer a variety of cluster technologies, the only netscaler that can really simplify management and realize intelligent traffic scheduling is the most unique.
Before we discuss cluster technology, let's talk about load balancing high-availability technologies:
High-availability technologies include these big categories
Master/Standby: The common scenario is that two ADCs are deployed in two machines, one at a time, and the other in hot standby. Disadvantage is a bit of waste of resources, do not support the simultaneous failure of more than 2 units.
Master/Master: VRRP-based technology, relatively few scenarios. If it is still 2 devices, but in the heart feel the equipment utilization increased, but the throughput of each can still not exceed 50%, or the failure of a machine will appear discarded business situation. If more than one can do so-called m:n, but the configuration is complex.
GSLB: Site-based high availability does not resolve high-availability within sites.
Cluster: Emerging high-availability technology.
First look at what is a cluster? This is very simple, not clear to ask a degree niang. The focus of today's discussion is on NetScaler clusters.
NetScaler transformed the architecture into Ncore in 2007, although it was still a traditional HA deployment model, but it has been transformed into a solid technical foundation for future cluster technology leadership. NetScaler's cluster comes from its own ncore architecture, which implements cluster based on full software, rather than half-device semi-software mode. It is thanks to this radical architectural transformation that NetScaler can build two cluster of L2 and L3, the difference is only for the NetScaler itself, and the client and the server does not have a strict network ownership relationship.
What is NetScaler Cluster?
Cluster is a solution to the redundancy scheme at the device level, overcoming the disadvantage of low equipment utilization under HA mechanism, uneven scale-out and the inability of the service to be redundant under special requirements.
Features of Citrix cluster:
o Single IP Management
o Single Business VIP
o All or part of the equipment involved in the calculation
o Flexible grouping and control
Typical cluster topology diagram
Custer has many concepts to understand:
Cluster type of Address
O Stripped
- Striped address, which belongs to all node in the cluster instance
O Partially Stripped
- Part of the spoke address, which belongs to a group of node in the cluster instance
O Spotted
- Node address, which belongs to one node in the cluster instance
Node state
o Active
- Synchronize cluster information and process business. The State of node at work
o Spare
- Synchronizes cluster information, does not process business, and can become active at any time. Often used to make hot preparation
o Passive
- Synchronizes cluster information, does not process business, and the default state of new node. Often used to have a plan to get node offline
What is Clag
o Cluster Link Aggregation
o Cross-device link aggregation
L2 Cluster
2-tier cluster, requiring each node to be in the same 2-tier network
L3 Cluster
3-layer cluster, each node can be in a different 3-tier network, that can cross the data center
Cluster deployment Scenario
The value of cluster
1, improve business reliability, improve equipment utilization
2, weaken the pre-business planning, protection of investment
3, simplified management, flexible scheduling
Comparison of cluster schemes
Comparison with friends and business
Case:
Summarize:
1, can customer service ha mode of low resource utilization, dual-machine failure problems
2, can provide high scalability, easily scale to the maximum of 32 6.4T throughput
3, can achieve 6 large scene different requirements of high availability
4, resource pooling, protection of investment, reduce the difficulty of operation and maintenance.
The king-netscaler Cluster of load Balancing cluster technology