SOS Facilitates Analysis

SOS has a high ingest rate

Many databases' ingest rates cannot keep up with large, high fidelity data. SOS was developed to handle such data, such as from HPC monitoring data, which may have thousands of metrics from tens of thousands of nodes, per second.

SOS supports efficient analysis in python

SOS, and its numSOS interface, can present query results as zero-copy ndarrays, which facilitate analysis, such as those over components and over time, as would be common in HPC monitoring data analysis. Other typical python-based analysis tools (e.g., jupyter notebooks, SciKit Learn etc) can also interact with SOS via python and ipython interfaces and return data structures.

SOS has command line tools for examining databases

SOS supports long-term analysis, through sharding