Metadata scaling limitations arise from the use of a centralized metadata server, also referred to as a metadata controller (MDC) or from other architectures that are unable to take advantage of additional compute and storage resources. Poor metadata scalability can result in degraded performance, even for routine tasks. Single server architectures limit the computational resources available for generating the massive number of IOPS needed to read, write, and search large volumes of metadata. With file and object populations in single data collections sometimes exceeding one billion items, even a single search that requires walking the full namespace might take many hours or days. Similarly, administrative functions like metadata dumps and restores can require extended periods of time.
