I'm trying to draw a database design of an ecommerce, and fulfilment of order platform. The company currently has a distribution centre for fulfilling the orders. But they want to extend this to use its stores for a part of the fulfilment process. I have designed a database of "internet sales" and "store sales", but I am stuck on the fulfilment of the internet order, and I wonder if any of you can help me with this.
Scenario : When the customer places in an order, and the distribution centre doesn't have a stock of an item to ship to the customers, the item needs to be taken from one of the stores. This item is then sent to the customer.
But the problem is that I can't just take an item from a store, and then send it to the customers, because the item hasn't been sold in the store, its (store) stock database isn't going to be updated. If I put the item through the cash machine, the item is removed from the stock table, but there are two transactions for the same item - one transaction from the internet, and the other from the store.
I guess my question is, how do I go about processing internet orders, and avoid having two transactions on the same item?
Any helpful pointers on this issue is greatly appreciated.
Update : Here's what I have done so far after advice from Jo Douglass,
Database Design Here
Sorry, I can't post images, because I don't have enough points. And please note that the above database design isn't complete
It sounds like you have a Transaction entity, and you have or are planning on having some logic which ensures that when one of these is created for an Item, your system knows to deplete the stock level for the relevant location (either a store of the distribution centre).
You could use an entity which shows an Item being transferred from one location (a store) to another (a distribution centre), and then create some logic which works very similarly to your existing logic - depleting the stock level in the starting location, and increasing the stock level in your destination location. Then when you carry out the last part of the process (sending the item to the customer), you'll have a Transaction showing that and depleting the distribution centre's stock level. Depending on the rest of your model, you might carry this out via a change to the Transaction entity, or by creating a new entity altogether.
Alternatively, if that doesn't really model what's happening in the business very well, then maybe you just need to modify your logic (and possibly your model - hard to tell without seeing your existing model). Rather than only being able to create store transactions via use of the cash register, perhaps you simply need to be able to create a store transaction that's been kicked off via the Internet.
One idea is to go ahead and treat the item as sold from the store (through an online transaction) and credit the store's account with the sale price. The distributor has probably already received the wholesale price from the store so it's happy, the store gets credit for the sale (with at least some part of the shipping charges) so it's happy, and you don't have to create new transaction codes or any other modification to the existing database.
Related
Okay, my partners and myself created these a while ago. We are going to be transferring this into SQL through Visual Basic soon, but I want to make sure everything is ready to go. Two major complaints that we were not able to fix was...
"By having Transaction and Product directly connected you are unable to allow multiple products on the same order (customer can't order both a latte and cappuccino on the same order)."
"Membership table: Not sure what the data in DiscountTypeTotal means - do you have multiple pieces of data in the same field? (then your table isn't on 1NF).
It looks like you need to allow each member to have multiple discounts - so you need another table to capture that."
How do we effectively correct these? How else would we connect Transaction with Products? I understand that the customer can only purchase one item per transactions, so would we have products and another table for multiple items? Allowing multiple customers to have discounts, I am lost. Any help would be appreciated.
Transaction is an object which may describe a recurring transaction (an object of it "says" such things as "we receive a $10 payment every month from this PayPal user").
Organization is a customer which may pay us and in return receive some services. Currently one organization can receive max one service, but this may change in the future.
PricingPlan is something like "Gold $20/month".
Purchase links together a pricing plan with a transaction.
Note that we cannot (or can but with a hardship) add new fields to Transaction model, because it is a part of a reusable module. The rest three models are under our complete control.
I need to make a relational DB structure from these four models.
I've drawn two possible relations between the models. Which of the two: the top one or the bottom one is more normal than the other?
The title is somewhat hard to understand, so here is the explanation:
I am building a system, that deals with retail transactions. Meaning - purchases. I have a database with products, where each product has an ID, that is also known to the POS system. When a customer makes a purchase, the data is sent to the back-end for parsing, and is saved. Now everything is fine and dandy, until there are changes to the products name, since my client wants to see the name of the product, as it was purchased then.
How do I save this data, while also keeping a nice, normal-formed database?
Solutions I could think of are:
De-normalization, where we correlate the incoming data with the info we have in the database, and then save only the final text values, not id's.
Versioning, where we keep multiple versions of every product, and save the transactions with the id of the products version, when it came in. The problem with this one is, that as our retail store chain grows, and there are more and more changes happening to the products, the complexity of the whole product will greatly increase.
Any thoughts on this?
This is called a slowly changing dimension.
Either solution that you mention works. My preference is the second, versioning. I would have a product table that has an effdate and enddate on the record. You can easily find the current record (where enddate is null) or the record at any point in time.
The first method always strikes me as more "quick-and-dirty", but it also works. It just gets cumbersome when you have more fields and more objects you are trying to track. It does, in general though, win on performance.
If the name has to be the name as it was originally, the easiest, simplest and most reliable way to do that is to save the name of the product in the invoice line item record.
You should still link to the product with a ProductID, of course.
If you want to keep a history of name changes, you can do that in a separate table if you wish:
ProductNameID
ProductID
Date
Description
And store a ProductNameID with the invoice line item.
I currently have an MS Access application that stores information about the employees who have responsibility for a certain task.
My form goes like this. You enter the task in a textbox. You pick the employee from a combo box which is bound to the staff table. And finally we save this information to the Task table. As simple as that I thought...
But here is the problem. No employee works forever in a company. A new/another employee maybe assigned the task which was previously carried out by an employee who is no longer working there. In the form once I update the new employee the old employee information is replaced and we wouldn't even know that that employee existed.
I came across the concept of adding effective dates to the employee which may be used to track the history of employees.
Now I would like to know how I would be able to preserve the data of the previous employee when I update the Task form with the information of the new employee.
What should I do?
Thank you for your assistance.
BR,
Paul
I have uploaded links to three forms that I am working on.
In the outside company officer form we don't have the date fields mentioned.
In our companies proposal form we might need to add additional officers but don't want to repeat the proposal information just the officer, division, alternate officer & division.
In the outside company detail form we might need to apply the same concept to the director of the company without applying to other institute information.
I was also researching on this topic where I found a document which explains the concept of effective dates. I need to know how to apply this and also without making a lot of changes to the application.
www.gsa.gov/graphics/staffoffices/DatedInformationandDateTracking.pdf
I hope this helps you understand my problem.
It would seem you are keeping very simple information, you want to know who is currently assigned to any given task. Add a "assignedDate" field to your table. Then you can change any reports or forms where you need to display the current assignee to a query whereby you group by "task" and select only the latest record.
Do not include the assignee as a field in the task table. You would need an additional table, say "TaskAssignees" comprised of TaskID, ContactID, entry date, assigned date, unassigned date. Now in reporting you can pull the correct assignee given any date or list a history of the assignees from start to finish.
I would say that you are making the issue too complicated.
Surely there's an administrative policy on what happens to tasks when an employee leaves -- their replacement takes them over or they are assigned to the departing employee's supervisor or a co-worker.
Thus, your application should encode that business practice. The form where you edit the employees should be set up so that when an employee leaves, their tasks are moved to the appropriate employee. This means you don't have to store effective dates and then write the complex SQL to filter by those dates. Instead, you'll use the same structure as you already have.
Now, of course, I'm not stupid -- I know that many companies SHOULD have such policies, but very often, these things fall through the cracks. When I'm in this position (as an outside contractor) I tell them their two options and price the two options. Implementing a policy for what happens with the assigned tasks when the employee leaves is cheap. Implementing an effective date is EXPENSIVE. So, if they want to save money, they'll come up with a policy that can be implemented in the database application.
If the solutions already suggested don't work for you then you could look at creating an audit table which effectively logs every change that is made to a record (This would have the advantage of allowing you to track ALL changes, even in areas of your system you have not yet developed).
Here are a couple of links that show how you could go about doing this:
a simple solution for tracking changes to access data
Allen Browne - creating an audit log.
What is the general pattern or approach to managing transactions in a web-based e-commerce system? How do you handle the situation where more than one user tries to buy the last item, for example?
To prevent two users from purchasing the same stock item of which there is only 1 unit in stock, you need to check that each item in a user's cart has stock available right before you create an order and decrement stock for that item.
This operation will have to be atomic and only one order can be processed at any given time (read: database transaction), which should not be a problem if you are using a central database for stock management.
If stock has run out by the time a client checks out, you should remove the item from the client's cart and redirect them to their shopping cart, informing them of the situation.
Of course, this situation only occurs when two users both add the same stock item to their cart of which only one unit is in stock and one of them checks out. First come, first served. You should generally not allow clients to add products to their cart if no stock is available at that moment, unless you can order new stock within a reasonable amount of time, but in that case, the whole point is moot.
You can take a preemptive approach by checking that stock is available the moment a client initiates checkout and take the same route as above. However, that would depend on the nature of your product and the volume of transactions vs cancelled orders. If it is likely that another order for the same item was cancelled in the meantime and stock becomes available by the time a client checks out, then you don't want to lose a sale by telling the client that no stock is available. Better to let the order fail at the moment stock is not available and inform the client of the situation, which is rare after all.
Why not take the order and then get that item for the customer, perhaps a bit later? You can win a repeat customer :)