Information Technology Reference
In-Depth Information
the low latency in all processes and the capability to support a very high rate of
transactions.
In 1999 eBay faced a series of service disruptions. In particular, over three days,
overloaded servers shut down temporarily without warning, meaning users couldn't
check auctions, place bids or complete transactions during that period. This led the
company to re-design its IT infrastructure, rebuilding its data centers according to
a Grid-type architecture in order to achieve a more flexible, scalable and reliable
infrastructure (Gil 2009).
The eBay Grid infrastructure consists of many small servers supported by some
higher-processor-count servers for a federation of back-end databases. eBay can
actually run on as few as 50 servers, which can be Web servers, application servers
and data-storage systems. Each of these servers runs separately, but communicates
with the others, thus each of them is notified if there is a problem in the network.
Growth can easily be achieved by adding servers to the Grid accordingly to demand.
Despite that, an infrastructure of only fifty servers is quite adequate in order to
run the site, eBay has one hundred and fifty servers more, in sets of fifty, in three
different locations, which are spread all over the world. These servers store the same
data, so if the main system crushes there are three other mirror systems to pick up
the slack. This new architecture based on Grid allows very high fault tolerance, the
elimination of the single point of failure, and easy growth together with low oper-
ating costs (Gibson 2004, MacFarland 2006).
eBay's business case is a representative example of a company that utilizes an
internal Grid solution to improve the utilization of its resources. Its revenues are
not originated from the Grid itself but from other sources (auctions, advertizing
etc). However Grid is used to increase the performance of the operation while also
reducing its expenses. Interconnecting all those servers in a Grid attains the afore-
mentioned performance enhancements, the high exploitation of resources, cost-effi-
ciency and economies of scale, due to the fact that interconnection of all machines
improves the utilization of each one. The main actor in this business case is the
company itself since it constructed its own Grid.
Table 5.3:
Example of Grid business case 1: The eBay
Grid Business Case
The Grid benefit (value-propo-
sition or added-value)
Main actors involved
1. Company utilises an internal
Grid solution as a virtualisation
technique to improve the utilisa-
tion of resources (also known as the
“Enterprise Grid”):
eBay
For the company: Performance
differentiation, cost-efficiency,
high exploitation of resources.
The company:
eBay
Grid s/w and application
providers:
eBay
