I've been working with the relationship-improvements branch of ember-data and I like the changes. However, I notice it's quite a bit slower than creating native ember objects.
- With Chrome, loading/finding 400 records is 3x slower with ember-data than if I create native ember objects (700ms vs 200ms).
- With IE8, it's almost 4x slower (2300ms to 600ms).
I'm wondering if the architecture of ember-data prevents it from being suitable to large data sets. Or if it's slow simply because it's young and we can expect that, over time, its performance will approach native ember-objects.
Given I have very large data sets in my app (1000+ records loaded (though not all visible at once)), I'm wondering if, it's better to just role my own much simpler data library or hang in there with ember-data.
We're launch our app in the next 1.5 months, and my guess is ember-data won't become dramatically faster by then.