a lot of changes
diff --git a/wally/suits/io/agent.py b/wally/suits/io/agent.py
index f6c3308..57ba229 100644
--- a/wally/suits/io/agent.py
+++ b/wally/suits/io/agent.py
@@ -293,15 +293,24 @@
 
 
 def slice_config(sec_iter, runcycle=None, max_jobs=1000,
-                 soft_runcycle=None):
+                 soft_runcycle=None, split_on_names=False):
     jcount = 0
     runtime = 0
     curr_slice = []
     prev_name = None
 
     for pos, sec in enumerate(sec_iter):
-        if soft_runcycle is not None and prev_name != sec.name:
-            if runtime > soft_runcycle:
+
+        if prev_name is not None:
+            split_here = False
+
+            if soft_runcycle is not None and prev_name != sec.name:
+                split_here = (runtime > soft_runcycle)
+
+            if split_on_names and prev_name != sec.name:
+                split_here = True
+
+            if split_here:
                 yield curr_slice
                 curr_slice = []
                 runtime = 0
diff --git a/wally/suits/io/ceph.cfg b/wally/suits/io/ceph.cfg
index 5593181..a10adfb 100644
--- a/wally/suits/io/ceph.cfg
+++ b/wally/suits/io/ceph.cfg
@@ -1,12 +1,16 @@
 [defaults]
-wait_for_previous
-group_reporting
-time_based
+wait_for_previous=1
+group_reporting=1
+time_based=1
 buffered=0
 iodepth=1
 softrandommap=1
 filename={FILENAME}
 NUM_ROUNDS=7
+thread=1
+
+# this is critical for correct results in multy-node run
+randrepeat=0
 
 NUMJOBS={% 1, 5, 10, 15, 40 %}
 NUMJOBS_SHORT={% 1, 2, 3, 10 %}
diff --git a/wally/suits/io/rrd.cfg b/wally/suits/io/rrd.cfg
index 5593181..3f4c074 100644
--- a/wally/suits/io/rrd.cfg
+++ b/wally/suits/io/rrd.cfg
@@ -6,50 +6,19 @@
 iodepth=1
 softrandommap=1
 filename={FILENAME}
-NUM_ROUNDS=7
-
-NUMJOBS={% 1, 5, 10, 15, 40 %}
-NUMJOBS_SHORT={% 1, 2, 3, 10 %}
+NUM_ROUNDS=35
 
 size=30G
 ramp_time=15
 runtime=60
 
 # ---------------------------------------------------------------------
-# check different thread count, sync mode. (latency, iops) = func(th_count)
-# ---------------------------------------------------------------------
-[ceph_test_{TEST_SUMM} * {NUM_ROUNDS}]
-blocksize=4k
-rw=randwrite
-sync=1
-numjobs={NUMJOBS}
-
-# ---------------------------------------------------------------------
-# direct write
-# ---------------------------------------------------------------------
-[ceph_test_{TEST_SUMM} * {NUM_ROUNDS}]
-blocksize=4k
-rw=randwrite
-direct=1
-numjobs=1
-
-# ---------------------------------------------------------------------
 # check different thread count, direct read mode. (latency, iops) = func(th_count)
 # also check iops for randread
 # ---------------------------------------------------------------------
-[ceph_test_{TEST_SUMM} * {NUM_ROUNDS}]
+[rrd_test_{TEST_SUMM} * {NUM_ROUNDS}]
 blocksize=4k
 rw=randread
 direct=1
-numjobs={NUMJOBS}
-
-# ---------------------------------------------------------------------
-# this is essentially sequential write/read operations
-# we can't use sequential with numjobs > 1 due to caching and block merging
-# ---------------------------------------------------------------------
-[ceph_test_{TEST_SUMM} * {NUM_ROUNDS}]
-blocksize=16m
-rw={% randread, randwrite %}
-direct=1
-numjobs={NUMJOBS_SHORT}
+numjobs=5
 
diff --git a/wally/suits/io/verify.cfg b/wally/suits/io/verify.cfg
index 4a66aac..58b8450 100644
--- a/wally/suits/io/verify.cfg
+++ b/wally/suits/io/verify.cfg
@@ -8,31 +8,21 @@
 filename={FILENAME}
 NUM_ROUNDS=1
 
-size=5G
+# this is critical for correct results in multy-node run
+randrepeat=0
+
+size=50G
 ramp_time=5
-runtime=360
+runtime=60
 
 # ---------------------------------------------------------------------
-# check different thread count, sync mode. (latency, iops) = func(th_count)
+[verify_{TEST_SUMM} * {NUM_ROUNDS}]
+blocksize=4k
+rw=randwrite
+direct=1
+
 # ---------------------------------------------------------------------
-[verify_{TEST_SUMM}]
-blocksize=4m
+[verify_{TEST_SUMM} * {NUM_ROUNDS}]
+blocksize=4k
 rw=randread
 direct=1
-numjobs=5
-
-# ---------------------------------------------------------------------
-# check different thread count, sync mode. (latency, iops) = func(th_count)
-# ---------------------------------------------------------------------
-# [verify_{TEST_SUMM}]
-# blocksize=4k
-# rw=randwrite
-# direct=1
-
-# ---------------------------------------------------------------------
-# direct write
-# ---------------------------------------------------------------------
-# [verify_{TEST_SUMM}]
-# blocksize=4k
-# rw=randread
-# direct=1