Improve performance for sequential read from big volume
During tests OpenVDS with big volume (like 56Gb) we found serious performance degradations (up to 3x).
Profiling in Intel VTune show problem in VolumeDataPageAccessorImpl::m_pages. Changing container type from std::list to std::unordered_map help as to resolve this:
std::unordered_map<int64_t, VolumeDataPageImpl*> m_pages;
Where key is chunk index.