The two central concerns for any network administrator who is tasked with the responsibility to make sure databases are working, accessible and providing information in a timely manner are “how much” and “how fast.”
Some might argue one is a subset of the other, but that viewpoint doesn’t take software into account. Networks are undoubtedly reliant on their hardware, but like a race car, the hardware is only as useful as the software allows it to be. Likewise, the fastest car on Earth will win few races with an inexperienced driver.
The Amount of Big Data
A network platform which occasionally requires employee access (access that should only be granted to employees who understand the system and have successfully completed an employee training and development program) to a very large number of records is what we call a “record heavy” database. It might be a financial institution, for example, that settles all its accounts once per day, but that must process all of the information in its database every 24 hours.
Such an application is much better modeled by a delivery truck than a conveyor belt. Since settlements can’t take place until all of the records have been entered, the application must be ready to move a large volume of data at once. For a network administrator and the hardware involved, this means large mass storage, large amounts of RAM and wide pipes.
The Frequency of Big Data
Let’s suppose another network administrator is tasked with setting up a database that requires continuous access to a large set of database records. This might be a library, for example, or a retail catalog where customers and users are continuously trying to locate database records, but are only looking for a handful of records at a time.
Such a network would be far less concerned about bulk and far more concerned about agility and automation. The CPUs and the routers in such a network would be required to move quickly and efficiently from one request to another in an environment where large amounts of RAM and mass storage (while still valuable) won’t have as much of an effect on the effectiveness of the network. This still qualifies as big data but it also qualifies as smart data. In this kind of network, the firmware and software would have to be optimized to prevent a situation where there are too many requests and too few resources to handle them.
To increase agility, flexibility and efficiency, a network administrator would have to concentrate their efforts on simplifying the system as opposed to adding more complex software and topology.
Reliable
Neither type of big data platform will function at its top potential if it is unavailable to its users. There are many ways to improve database reliability. Among these techniques network administrators will find useful knowledge and tools they can use to make their systems available more often to more users.
One of the keys is found in the strategy for agility-heavy networks like the one in the library and retail catalog examples. Such networks are universally effective in big data applications because they are only limited by their capacities in delivering the functionality users need. Multiple such networks employing the principle of redundancy can rapidly scale to meet the needs of many different kinds of applications in ways that record-heavy databases can’t.
The principles of database management dont necessarily change because of the volume of data or the number of users. This is especially true for any company that makes employee training and development a priority. Effective and properly implemented database and network structures will serve users well as the size and speed of the databases change, in much the same way the basics of pitching a baseball doesnt change just because a baseball player is in the World Series as opposed to a high school playoff game.
What matters most is investing the time and effort necessary to learn those principles and when to put them into practice.
