Merge pull request #616 from Rajadeepan/doc

Add Kubemark Benchmark Design Doc

Author: k8s-ci-robot

doc/design/Benchmark/kubemark/kubemark-benchmarking.md

@@ -0,0 +1,43 @@
+## BACKGROUND
+
+Performance evaluation of KubeBatch scheduler. Kubemark will be used as a tool for benchmarking.
+Kubemark is a performance testing tool which we use to run simulated kubernetes clusters. The primary use case is scalability 
+testing, as creating simulated clusters is faster and requires less resources than creating real ones.
+
+## OBJECTIVE
+
+-  Evaluating scheduler performance of kube-batch
+
+## DESIGN OVERVIEW
+
+We assume that we want to benchmark a test T with kube-batch scheduler.
+For the benchmarking to be meaningful,these kube-batch should be running in a kubemark cluster and
+at  scale (eg.  run 100 nodes).
+At a high-level, the kubemark should:
+
+- *choose the set of runs* of tests T executed on environment using the kube-batch scheduler by setting "schedulerName: kube-batch" in the pod spec,
+- *obtain the relevant metrics (E2e Pod startup latencies)* for the runs chosen for comparison,
+- *compute the similarity* for  metric, across both the samples,
+
+Final output of the tool will be the answer to the question performance of the kube-batch scheduler. 
+
+
+
+### Choosing set of metrics to compare
+
+Kubermark give the following metrics
+- pod startup latency
+- api request serving latency (split by resource and verb)
+
+In our case we would requiring <b> E2E Pod Starup Latency </b> which would include
+   - The Latency for <b> scheduling the pod (create to schedule) </b>
+   - The Latency for <b> binding the pod to node (schedule to run) </b>
+
+## Changes to be done to the kubemark
+
+  1) Modify the kube-mark startup script to bring up kube-batch scheduler along with the external kubernetes cluster. <br /> 
+     Create CRDs for podgroup and queue.
+  2) Run the density/latency test 3k pods on 100 (fake) nodes with batch scheduler.<br />
+     For Batch scheduler test case : Modify the density/latency test case to create podgroup and add the podgroup name to the annotation along with "schedulerName: kube-batch" in the pod spec.
+  3) Print out the metrics.
+