NFS sponsored a data storage visioning workshop in 2018. The workshop was hosted by IBM Research. From the workshop, they produced a report, summarizing the discussion from the workshop. The report outlines the future research challenges and opportunities that the attendees recommend to work on. The discussion of the workshop was organized in the four groups. Below are some interesting points I copied from the report. Storage for the Cloud, Edge, and IoT Systems More desegregated, composable software architectures are highly desirable.

Quite often, we need to know the network round-trip time (RTT) between two nodes and a fairly well known tool is ping. However, I found out the proper way to use the ping tool only until today. Actually, I guess I should just use the netperf tool instead, to measure the network latency in the future. By default, ping sends out a packet every second. The interval which ping uses to send out a packet can have a huge impact on the measured network latency.

I really liked the paper. The idea is neat. The paper is also well written and the design and the new idea is clearly presented. A very good paper to read and learn from on how to write good papers. Summary The key idea is to convert random small writes into large sequential writes at the device driver. This reduces data fragmentation and also reduces request handling overhead for the SSD, as it can combine multiple small write requests into fewer large write requests.

Summary Copysets proposed another approach/perspective for data replication, than the traditional random replication. In random replication, N nodes are selected randomly from the cluster to store a data chunk. Random replication provides a strong protection against uncorrelated failures. For each individual chunk, since it can be stored randomly at N nodes in the cluster, it is quite resilient to data loss. However, when we consider all data chunks, any N node failure will almost lead to loss of some chunks, as long as these chunks happen to be replicated at these exact N nodes.

How time flies! I am 33 years old now and have lived in US for more than 10 years. I have also graduated from University of Utah for almost 5 years now. Wahoo! Each week passed very quickly. Now is the time to pause a little bit and reflect what has been achieved and what could be done better. Achievements Worked on a few projects/papers and made a few submissions to FAST/ATC/HotStorage, though none of them have been accepted yet.

This blog lists tools/simulators/platforms people used to evaluate SSD researches. MQSim MQSIM FAST18, ATC19-Kim SSDSim SSDSim Microsoft Patch [ATC12][atc12] FEMU FEMU-fast18-short