第3届Mathorcup数学建模竞赛优秀论文-ProblemA(English)-ProblemA.doc
版权申诉
140 浏览量
2024-03-21
21:51:53
上传
评论
收藏 25KB DOC 举报
Train ticket website optimization
Trains in China is an important means of transport, especially in the Chinese Lunar New Year
travelers buy tickets will be very much able to buy a train ticket is every traveler's desire. Since
the end of 2011 after the implementation of network tickets, buy tickets way train ticket from
the mouth only to queuing at the ticket, and gradually developed into a website or via telephone
booking tickets through a variety of convenient ways. Online ticket really do give to the
convenience, tickets could not see the house on the computer will be able to buy tickets, the
ticket window would not have so busy, saving a lot of manpower and resources. However, in
2013 Spring Festival ticket website when there is a large number of issues, login, purchase,
payment and other aspects have had problems, to bring trouble ticket buyers.
On the website booking process to make the following simplification:
1 A query I vote for this lookup database trains remaining tickets
2 to generate orders, locked remaining tickets for the purchase of a ticket, according to the ticket
information and ticket information generation who paid orders
3, completion of the payment, the ticket purchase is complete; payment fails, ticket booking
recycling continues to others
Analysis of the site where the bottlenecks are:
The first site traffic problems, visit the Web site at the same time the number of excessive
server was unable respective owners request, resulting in not logged in, can not query the ticket
or unable to pay more. During the Spring Festival traffic to your site is very large, the site's daily
visits can refer alexa records (see Resources). In order to alleviate the problem of excessive same
time to visit the site to take a timeshare purchase method, equivalent to the amount of
dispersion of the request, reducing at the same time the volume of traffic problems.
The second problem is the only resource that is a ticket booking people first of all to lock a
ticket, if you have too many people at the same time to rob the tickets, then the system will not
be able to determine that this ticket Who locked, so that the system will make a request to retry,
has been unable to achieve lock busy booking tickets will cause the system to fail. In response to
this ticket only resource contention, the site has taken a queuing method, that request comes
first obtain a number of holding a number waiting queue, thus avoiding a direct result of the
deadlock on the ticket, get a number that is very fast , while the processing speed relative to a
ticket to take a number, into the queue is relatively slow, because there are locked tickets,
generate orders, testing a series of operations to pay. (This process can be compared and the
bank's operating procedures, to obtain a first number of queued for processing, the service
window, a processor, in accordance with the FIFO queue, the order processing).
Booking sites you want to utilize a queue to meet the current ticketing needs, annex gives
the queue test data: a single server test data into the queue and a server on the queue data.
Which enqueuing data sampling rate is 1%, out of a queue is a queue of all the data.
Answer the following questions based on the data
1, evaluate existing timeshare ticket policy is reasonable, how you can optimize (requires no
more than 5 timesharing)
资源评论
