How much data is too much for an array?

I have a flex webapp that retrieves some names & addresses from a database. Project works fine but I'd like to make it faster. Instead of making a call to the database for each name request, I could pre-load all names into an array & filter the array when the user makes a request. Before I go do开发者_如何学运维wn this route though I wanted to check if it is even feasible to have an application w/ 50,000 or 1 million elements in an array? What is the limit b/f it slows down the app? (I anticipate that it will have a lot to do w/ what else is going on in my app but for this sake lets just assume the app ONLY consists of this huge array).

Searching through a large array can be slower than necessary, particularly if you're talking about 1 million records.

Can you split it into a few still-large-but-smaller arrays? If you're always searching by account number, then divide them up based on the first digit or two digits.

To directly answer your question though, pure AS3 processing of a 50,000 element array should be fine. Once you get over 250,000 I'd think you need to break it up.

Displaying that many UI elements is different though. If you try to bind a chart to a dataProvider with 10,000 elements, it's too much. Same for a list or datagrid.

But for pure model data, not ui bound, I'd recommend up to 250,000 in my experience.

If your loading large amounts of data (not sure if your using Lists though), you could check out James Wards post about using AsyncListView with paging to grab the data in chuncks as its needed. Gonna try and implement something like this soon. His runnable example uses 100,000 rows with paging of 100 (works with HttpService/AMF type calls): http://www.jamesward.com/2010/10/11/data-paging-in-flex-4/

Yes, you could probably stuff a few million items in an array if you wanted to, and the Flash player wouldn't yell at you. But do you really want to?

Is the application going to take longer to start if it has to download the entire database locally before being able to work? If the additional time needed to download that much data isn't significant, are a few database lookups really worth optimizing?

If you have a good use case to do this, you're going to have to pay attention to the way you use those data structures. Looping over the array to find an item is going to be a bit slow, so you'll want to create indexes locally, most likely by using a few hash structures. The more flexible you allow the search queries to be, the more interesting the indexing issues will be.