Modern computers are often significantly impeded by issues stemming from memory access. As most systems use disk for large quantities of storage, there is a penalty for both random read latency and power consumption. In a distributed system, this means that accessing data sorted by a hashed key can be a bottleneck for client operations. Additionally, the wattage and price required to power a large server rack can be discouraging to those who want to host their own servers. Using low-power, diskless nodes, like raspberry pis (running with ARM 700 MHz CPUs), many of these problems can be solved. By running a large quantity of small nodes in parallel, a memory bandwidth comparable to a typical server rack can be achieved at only a fraction of the cost.