We got to the third and final article on high availability questions, I feel if you made it very long, but I've summed it up in such a way that neither is excessively long each item.
More high availability for all
The strategies are more common for data protection?
Strategies more common for data protection are:
- Backups sent to another physical location of backup on a regular basis.
- Backup in the same cluster, to speed up recovery
- Backing up data on servers outside the company. It is usually done with a SAN network
- Real-time data replication and backup
That measures are applied to external providers for the protection of data of a company?
In companies, normally apply a series of measures for the speedy recovery of systems.
- Local copies are synchronized, such as RAID.
- Protective voltage, to avoid energy increases in equipment.
- System of uninterruptible power supplies (UPS) and generators for cuts and supply failures.
- Fire prevention, Antivirus systems and Access Control
What is the SPOF?
The SPOF (Single Point of Failure) is the single point of failure, i.e., is the moment in which a redundant network is left with a single active support.
Setting an example, in a cluster of two balancers, in the moment in which one fails, the other becomes the SPOF of rolling, i.e., if the latter fails, is completely cut the supply and seriously affect the functioning of the organization.
What is the FailOver?
FailOver is the ability to change node at the time when a service fails. For example, at home we could have fiber at home and if this fails, we have connected a Router 3 G which offers internet where the primary network fails.
What is the TakeOver?
The TakeOver is the process by which a failover is automatically. Please note that takes place before the node fails, so that it can thus be replaced or repaired before the end of its useful life by reducing negative impacts on the availability of the service.
What is the GiveAway?
A giveaway or switchover is forced to perform, usually, failover maintenance on the removed node.
What is the Splitbrain?
To manage servers and to verify if the values of high-availability and SLA offered being taken, the nodes are sent mutually control packages to see if other nodes are alive. The splitbrain happens when multiple nodes are created that the remainder is inactive, which is intended to make charge at the same time the work of the other creating a conflict. For this is a "witness" called quorum which is which indicates that in the event of conflict, the owner of the quorum will be which take charge of the service. This bug should be fixed by the administrators of the system in the configuration to prevent worse consequences.
What is HotSwap?
HotSwap or change in heat is the ability of a team having changes in your hardware while it is turned on. For example, if a team fails you hard drive and has the ability to change it where there is no fault in the system without turning it off, this equipment allows hotswap. Currently all systems allow this feature, it is only to ease to make it different.
What is bonding?
Bonding is the union of different network cards to treat them as a single, uniting its network speeds. For example, if we have a team with two Gigabit interfaces and connect them to the same network, we can generate a bond to logically join interfaces, with what the actual speed of the team would be 2 Gigabit. In this way, we increased the network SPOF, allowing high availability.
What is a cluster?
We have spoken many times of cluster and that we need them to provide high availability, but never explained that they are exactly…
A computer cluster refers to a set of computers that behave as a single team. Clusters can be classified according to their characteristics:
- Hight performance cluster computing or high performance cluster are clusters that are running with a great need for computational or memory tasks. This type of tasks often require quite some time in the cluster.
- High-availability cluster or hight availabilty cluster computing are clusters that provide availability above all, with failures and tolerant SPOF hardware detection software.
- Hight throughtput computing clusters or cluster of high efficiency are clusters in which its objective is the realization of most of the tasks in the shortest possible time.
What is a fault-tolerant load balancer?
A fault-tolerant load balancer is a system to provide redundancy and high availability of services making these act as one single. For example in a web service can have several teams back to offering the service and the load balancer will distribute the traffic according to various parameters, as the load system, types of client, etc.. Fault tolerance comes when a service node falls, with which the load balancer automatically resets requests to active servers to avoid service cuts that could reduce the SLA.
What is round robin?
Round Robin is an algorithm to select a group of items in the order of processing of these. Cluster, this is used to balance the networks dealing equitably through a protocol ignoring the State of the servers.
What is the LVS?
LVS stands for Linux Virtual Server, a high scalability and high-availability system built on a cluster of physical servers. The designed architecture is completely transparent to users and users feel that they are trying to by a single final server.
What is Heartbeat?
The heartbeat is a daemon that allows customers of clusters to know the status of the services thereof to so control that is taking a high-availability.