genXML.sh script to generate COSbench XML files for s3 benchmarking.
Also includes 'cbparser.py', a script which parses COSbench workload results archive directory
and produces text results file.
USAGE: ./cbparser.py -d archive/w25-empty-s3/
NOTE: genXML.sh generates a random string as container prefix name to avoid AWS S3 name collisions
Creates these XML files
FILENAME | Description |
---|---|
fill.xml | cluster fill workload (creates buckets and objects) must be run as first workload |
empty.xml | cluster empty (removes objects and buckets) must be run as last workload |
seqops.xml | performs sequential reads then writes. For writes it initializes new containers and perfoms write on them. Eventually it deletes and disposes these new containers and objects within it. |
randomops.xml | performs random reads then writes. For writes it initializes new containers and perfoms write on them. Eventually it deletes and disposes these new containers and objects within it. |
mixedops.xml | performs mixture of read, list, write, deletes |
Edit genXMLs.sh and set access key, secret key and endpoint
VARIABLE | Description |
---|---|
akey | s3 access key |
skey | s3 secret key |
endpt | s3 endpoint |
Further define workload conditions by defining these
VARIABLE | Description |
---|---|
testname | name prepended to all XML files |
runtime | workstage duration in seconds - not used in fill or empty workloads |
objSIZES | object sizes |
numCONT | number of containers/buckets |
numOBJ | number of objects (per bucket) |