1. Identify the risks to data integrity that eBay took when it transferred some of the DBMS functions, such as referential integrity, to the application programs.

2. What types of data required eBay to use both relational and NoSQL? Why does eBay use both of these?

VV full. Although the website is known best for its Inter- How eBay Scales Its Database Architecture with SQL and NoSQL provides, such as referential integrity, out of the database and into the ith more than 150 million users and nearly a billion items listed on its site, eBay IT staff have their hands application instead. For example, the DBMS could be configured to pre- vent someone from adding a new listing for an item if the item's seller did not yet have a record in the Seller table. The DBMS would check the net auctions, most of the items listed on the site are actually new. eBay earns revenue from the fees it charges sellers to list and sell their prod- Seller ID that the seller was trying to enter as a foreign key in the item record to ensure that the same Seller ID is in the Seller table. But that ucts on the digital commerce site and also from advertising. More than constraint adds workload to the database itself, and the same check half of eBay's revenue comes from outside the United States, making the could be accomplished in the application that sellers use to create list- platform truly worldwide. Like so many other successful tech companies, eBay continually faced the challenge of extremely fast growth. ings. Relieving the DBMS of that kind of work allows it to handle many more transactions and queries without slowing down. eBay's innovative auction platform was initially built in 1995 over a single weekend in a staff member's living room, combining cheap and eBay continues to develop strategies to handle growth, ensure secu- easily purchased parts from local stores. This architecture maxed out rity, and add new features, and the company has implemented NoSQL with only 50,000 items listed, so eBay implemented an Oracle database databases along with the relational models. The relational database is server for the data with additional servers to handle the applications. "tried and true" for tightly governing how the steps in a transaction are Initially, eBay was able to scale up the database server to handle bigger accomplished, ensuring security, and also recovering from glitches, espe- loads, but this approach also faced a limit to growth. cially those that might happen in the middle of a transaction. The NoSQL approaches, however, excel at handling extremely large data sets with less The next step was to scale out rather than up by splitting the database into shards so the data could be stored on multiple database servers. This structure, and processing speeds are extremely fast. For example, eBay provides a number of advantages, one of which is simply that more serv- uses Hadoop for big data analytics to better understand customer behav- ers could be added when needed to further share the database load. ior on the site, analyzing petabytes of customer clicks and page views. eBay engineers made further adjustments to relieve the load on Companies like eBay must constantly attend to changing business needs, the database servers by moving some functions that a DBMS ordinarily revamping their enterprise architecture as new technologies emerge