c# - Generic Object Cache -

i working on project plan on using redis persistent data storage, task @ hand, working on generic object cache. , huge fan of linq have started designing cache support this.

    public concurrentbag<object> cache = new concurrentbag<object>();      public list<t> findby<t>(func<t, bool> predicate) t : class     {         var tmp = new list<t>();          foreach (var in cache)         {             try             {                 t obj = t;                  if (obj != null)                     tmp.add(obj);             }             catch { }         }          return tmp.where(predicate).tolist();     } 

i afraid when object cache grows large become inefficient. (i estimate 500k-1m objects)

i hoping possible use this

    public concurrentbag<object> cache = new concurrentbag<object>();      public list<t> findby<t>(func<t, bool> predicate) t : class     {         return cache.where<t>(predicate).tolist();     } 

hopefully not off-track here? suggestions welcome :)

since estimate lot of items in cache , operations on cache type specific, use multiple bags wrapped dictionary. speed finding subset of cache of type of interest , ideal if cache contained many minimal subsets of different types.

readonly idictionary<type, concurrentbag<object>> _cache = new concurrentdictionary<type, concurrentbag<object>>();  public list<t> findby<t>(func<t, bool> predicate) t : class {     // check if items of type {t} exist in cache.     concurrentbag<object> bag;     if (_cache.trygetvalue(typeof (t), out bag))     {         // cast, apply predicate , return.         return bag.cast<t>().where(predicate).tolist();     }     // return empty list.     return new list<t>(); } 

of course need handle adding items cache ensure different types put corresponding bags.