diff --git a/CMakeLists.txt b/CMakeLists.txt
index a312719..46901a7 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -22,7 +22,7 @@ set(CMAKE_CXX_STANDARD_REQUIRED TRUE)
 string(REPLACE "-DNDEBUG" "" CMAKE_CXX_FLAGS_RELWITHDEBINFO "${CMAKE_CXX_FLAGS_RELWITHDEBINFO}")
 string(APPEND CMAKE_CXX_FLAGS_RELWITHDEBINFO " -g -DSPDLOG_ACTIVE_LEVEL=SPDLOG_LEVEL_DEBUG ")
 string(APPEND CMAKE_CXX_FLAGS_DEBUG " -DSPDLOG_ACTIVE_LEVEL=SPDLOG_LEVEL_DEBUG ")
-string(APPEND CMAKE_CXX_FLAGS " -Wall -Wextra")
+string(APPEND CMAKE_CXX_FLAGS " -Wall -Wextra ")
 
 ###
 # Mandatory: devices configuration file.
diff --git a/benchmarks/warm_benchmark.cpp b/benchmarks/warm_benchmark.cpp
index b323a18..d7a4a94 100644
--- a/benchmarks/warm_benchmark.cpp
+++ b/benchmarks/warm_benchmark.cpp
@@ -70,17 +70,25 @@ int main(int argc, char ** argv)
 
   // FIXME: move me to a memory allocator
 
-  rfaas::RdmaAllocator<rdmalib::Buffer<char> > rdmaAllocator(executor);
-  rdmalib::Buffer<char>* in = rdmaAllocator.allocate(opts.input_size);
-  rdmaAllocator.construct(in, IBV_ACCESS_LOCAL_WRITE);
-  rdmalib::Buffer<char>* out = rdmaAllocator.allocate(opts.input_size);
-  rdmaAllocator.construct(out, IBV_ACCESS_LOCAL_WRITE | IBV_ACCESS_REMOTE_WRITE);
+  // Sample: test demonstrating standard memory allocation.
   // rdmalib::Buffer<char> in(opts.input_size, rdmalib::functions::Submission::DATA_HEADER_SIZE), out(opts.input_size);
   // in.register_memory(executor._state.pd(), IBV_ACCESS_LOCAL_WRITE);
   // out.register_memory(executor._state.pd(), IBV_ACCESS_LOCAL_WRITE | IBV_ACCESS_REMOTE_WRITE);
 
-//  const rfaas::RdmaAllocator<rdmalib::Buffer<char> > &rdmaAllocator1 = rdmaAllocator;
-  std::vector<rdmalib::Buffer<char>, rfaas::RdmaAllocator<rdmalib::Buffer<char>>> v(8, rdmaAllocator);
+  // Sample: test demonstrating allocation with our custom allocator.
+  rfaas::RdmaInfo info_in(executor,IBV_ACCESS_LOCAL_WRITE,rdmalib::functions::Submission::DATA_HEADER_SIZE);
+  rfaas::RdmaAllocator<rdmalib::Buffer<char>> allocator_in{info_in};
+  rdmalib::Buffer<char>* in = allocator_in.allocate(opts.input_size);
+
+  rfaas::RdmaInfo info_out(executor,(IBV_ACCESS_LOCAL_WRITE| IBV_ACCESS_REMOTE_WRITE));
+  rfaas::RdmaAllocator<rdmalib::Buffer<char>> allocator_out{info_out};
+  rdmalib::Buffer<char>* out = allocator_out.allocate(opts.input_size);
+
+  // Sample: test demonstrating allocation with std::vector.
+  //  rfaas::RdmaInfo info_v(executor,(IBV_ACCESS_LOCAL_WRITE| IBV_ACCESS_REMOTE_WRITE));
+  //  rfaas::RdmaAllocator<rdmalib::Buffer<char>> allocator_v{info_v};
+  //  std::vector<rdmalib::Buffer<char>, rfaas::RdmaAllocator<rdmalib::Buffer<char>>> v(allocator_v);
+
 
   // TODO: Since the for loop writes a value of 1 to each byte of the in buffer,
   //       it overwrites all bytes previously set to 0 by the memset() function.
diff --git a/rfaas/include/rfaas/rdma_allocator.hpp b/rfaas/include/rfaas/rdma_allocator.hpp
index 78f8032..1691b03 100644
--- a/rfaas/include/rfaas/rdma_allocator.hpp
+++ b/rfaas/include/rfaas/rdma_allocator.hpp
@@ -11,12 +11,24 @@
 #include <rfaas/executor.hpp>
 
 namespace rfaas {
+
+  struct RdmaInfo {
+
+  public:
+    RdmaInfo(executor& executor, const int access, const int header_size=0)
+        : executor(executor), access(access), header_size(header_size) {}
+    const executor& executor;
+    const int access;
+    const int header_size = 0;
+  };
+
   template<typename T>
   class RdmaAllocator {
+
   public:
     typedef T value_type;
 
-    inline explicit RdmaAllocator(const executor &executor) noexcept: _executor(executor) {}
+    inline explicit RdmaAllocator(RdmaInfo& info) noexcept: _info(info) {}
 
     template<class U>
     constexpr RdmaAllocator(const RdmaAllocator<U> &) noexcept {}
@@ -26,36 +38,21 @@ namespace rfaas {
         throw std::bad_array_new_length();
 
       // Maybe we could directly call the memset function here
-      mmap(nullptr, size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, 0, 0);
-      if (auto buffer = static_cast<T*>(std::malloc(size * sizeof(T)))){
+      if (auto buffer = new rdmalib::Buffer<char>(size, _info.header_size)) {
         report(buffer, size);
+        buffer->register_memory(_info.executor._state.pd(), _info.access);
         return buffer;
       }
       throw std::bad_alloc();
     }
 
-    template <class U, class arg1>
-    void construct (U* p, arg1 access)
-    {
-      std::cout << "constructor" << std::endl;
-      p->register_memory(_executor._state.pd(), access);
-    }
-
-    template <class U, class arg1 , class arg2>
-    void construct (U* p, arg1 access, arg2 head)
-    {
-      std::cout << "constructor" << std::endl;
-      p->register_memory(_executor._state.pd(), access, head);
-    }
-//    [[nodiscard]] inline T *construct(const std::size_t &size, const int &access, int header = 0) {
-
     inline void deallocate(T *p, std::size_t size) noexcept {
       report(p, size, 0);
-      std::free(p);
+      p->~T();
     }
 
   private:
-    const executor &_executor;
+    const RdmaInfo &_info;
 
     void report(T *p, std::size_t n, bool alloc = true) const {
       std::cout << (alloc ? "Alloc: " : "Dealloc: ") << sizeof(T) * n