Merge of 2 active record arrays very slow?

I have 2 sets of data as following:

default_products - A set of active record objects retrieved from memcached owned_products - A set of active record objects retrieved from the database

Retrieving the data is fast and i get around 30-40 req/s when just returning either one of the arrays yet as soon as i do the following to return both:

all_products = default_products + owned_products

Performance plummets to around 3-4 req/s. Why is this operation so slow? Is there anoth开发者_开发百科er way to do this to speed things up?

The 'products' use STI to have subclasses such as 'Furniture' and 'Clothing' could this cause any performance problem?

Thanks

---

I'm guessing that using the '+' operator makes things slow because actually allocates a new object. I suppose this could also be the result of garbage collection being triggered. If you don't mind modifying one or the other array you can use the '<<' operator, which is really fast.

Here's how I measured the speed difference and illustrate the side effect of modifying the array.

require 'benchmark'

a, b = [], []
array_size = 1000
random_limit = 1000

array_size.times { a << rand; b << rand }

p "starting count and object_ids"
p a.count, a.object_id
p b.count, b.object_id

p "resulting object_ids"
p (a<<b).object_id
p (a + b).object_id


n = 50000
Benchmark.bm do |x|
  x.report { n.times {a + b }}
  x.report { n.times {a << b }}
end

p "notice that the count has changed from 1000 to 51000 for the '<<' method"
p a.count, a.object_id
p b.count, b.object_id

---

If you have no use for the old variables afterwards, you can use Array.concat

default_products.concat(owned_products)

That should be faster than adding the arrays together and returning the result as it doesn't have to create a new object.

See: http://www.ruby-doc.org/core-1.9.3/Array.html#method-i-concat