Clustering

Computer clustering has become very common in businesses that want to reduce downtime and increase high availability or load balancing.  There are many types of computer clusters out today, from file server, load balancing, database, web hosting, and even data processing.  Whether the cluster of servers are replications of one another or sharing the same task to reduce strain on one another, a cluster can be extremely useful in either sense.

 Server clustering is the concept of which you combine multiple instances of one specific type of server called a node for the means of high availability, load balancing, and even redundancy.  Each specific setup requires hardware communicating with one another at every given second to provide customers, employees, or clients with either load balancing or high availability and even a combination of both together. 

A database server cluster would consist of multiple servers with the exact same database installation on all of them.  The reason for this is to ensure the metadata repository is easily shared between the multiple nodes within the cluster.  Not only if, but when a fail-over does occur, there is another node within the cluster that is able to provide database functionality.  As long as the system is setup within an active-passive configuration which is basically only one server being active at any point in time and the others in a passive instance. 

If the database comes across some sort of issue or problem and fails only a slight, one of the passive instances can take on the task.  This would in turn provide the company with high availability and reduce downtime to null.  Even though this system would seem to be redundant, it does not actually provide redundancy but more or less a fail-safe high availability system.  I would suggest implementing a redundant drive just to be on the safe side.

In a real world example, if a casino's stores and gift shop wishes to provide twenty-four hour seven days a week service to its customers a database cluster server would suit their requirements best.  The reason, of which a database cluster of servers would be the best solution, is it provides high availability no matter the situation.  With this type of setup, at every point in time throughout the day and night, the database would be available.  

Even when considering high usage, load balancing would be able to handle anything you could throw at it, no matter how many people are accessing it at any given second the cluster of servers sharing the database would be able to produce the expected results.  The load balancing in combination with high availability cluster is especially useful when working at a casino gift shop as every second counts to produce extremely high profits.

A majority of the benefits available when working with clustered servers comes down to the ability to provide efficient services.  From providing these services, such as database access, at every second throughout the day using high availability, to being able to process a task that would take hours can take minutes or even seconds, which would be load balancing.  Even if the main database server were to fail the next node or server in the cluster would be able to pick up right where it left off.  When there are hundreds of customers trying to purchase products at a given point in time the database would be able to handle the load with the ability to balance the work load.

Challenges may consist of security, cooling, and housing of the clustered servers.  Security in itself is a bit of a challenge, staying up to date with every possible threat is nearly impossible.  One small hole in any part of the entire cluster security could bring the entire system down.  A management system could help keep the servers and its associated hardware updated and a key-card system for the facility of which they are housed.  Whether housing the server in house or out sourcing the cluster of servers, making sure they are in a secure facility is definitely high on the list of concerns. 

One main challenge that most people do not expect is the heat a cluster of servers could produce, which in turn requires cooling.  This is one of the main reasons people outsource their server hosting or housing to ensure the correct facility is in use.  As stated by Bob Ehlers, "Unfortunately for system designers, the relentless push for greater computing capacity has also resulted in greater heat being concentrated in a much smaller space (Ehlers, 2005)."  This is of course is the reason blade servers are not normally used in house but can be found in server farms or data-centers.  Tower servers allow the ability to cool the system easier in a facility onsite versus a blade server being used offsite.

Overall tower servers (COTS) can produce quite an efficient database cluster with others alongside it.  The reason for this is it allows you to secure, cool, and manage the cluster onsite versus outsourcing which in turn causes networking nightmares.  Even if the company is larger it would require multiple stronger servers to house onsite, but it would still get the job done.  Clustering servers may be a daunting task but the end result is far worth the benefits in the long run.

References

Ehlers, B. (2005, May). Cluster Computing and I/O: What Blade Servers Miss, CompactPCI Can Hit. Retrieved March 6, 2017, from RTC Magazine: http://rtcmagazine.com/articles/view/100339

Englander, I. (2014). The Architecture of Computer Hardware, Systems Software, & Networking An information technology approach. Don FowleY.

IBM. (n.d.). DB2 database clustering. Retrieved March 4, 2017, from IBM: https://www.ibm.com/support/knowledgecenter/en/SSZJPZ_11.3.0/com.ibm.swg.im.iis.productization.iisinfsv.ha.install.doc/topics/wsisinst_pln_ha_xmeta_db2.html

O'Reilly, J. (n.d.). Real pros and cons in the COTS server decision. Retrieved March 5, 2017, from Search Data Center: http://searchdatacenter.techtarget.com/opinion/Real-pros-and-cons-in-the-COTS-server-decision