diff --git a/lisa/analysis/frequency.py b/lisa/analysis/frequency.py
index a6c09a3135b7f6cc31151d1240681ae1bd75ba7e..97da69ef0f9003b3c125eb585b4b26de120c28a3 100644
--- a/lisa/analysis/frequency.py
+++ b/lisa/analysis/frequency.py
@@ -278,7 +278,7 @@ class FrequencyAnalysis(TraceAnalysisBase):
         :type clk: str
         """
 
-        logger = self.get_logger()
+        logger = self.logger
         start = self.trace.start
         end = self.trace.end
 
@@ -359,7 +359,7 @@ class FrequencyAnalysis(TraceAnalysisBase):
         If ``sched_overutilized`` events are available, the plots will also
         show the intervals of time where the system was overutilized.
         """
-        logger = self.get_logger()
+        logger = self.logger
 
         df = self.trace.df_events('cpu_frequency')
         df = df[df.cpu == cpu]
diff --git a/lisa/analysis/latency.py b/lisa/analysis/latency.py
index 9713c3fd2620dd056b44a24d83777dc2cfb494d2..a18b8ff4a8d18bf53387e1769cbe8bd6a6e792c9 100644
--- a/lisa/analysis/latency.py
+++ b/lisa/analysis/latency.py
@@ -196,14 +196,14 @@ class LatencyAnalysis(TraceAnalysisBase):
         if wakeup:
             df = self.df_latency_wakeup(task)
             if df.empty:
-                self.get_logger().warning("No data to plot for wakeups")
+                self.logger.warning("No data to plot for wakeups")
             else:
                 df.plot(ax=axis, style='+', label="Wakeup")
 
         if preempt:
             df = self.df_latency_preemption(task)
             if df.empty:
-                self.get_logger().warning("No data to plot for preemption")
+                self.logger.warning("No data to plot for preemption")
             else:
                 df.plot(ax=axis, style='+', label="Preemption")
 
diff --git a/lisa/analysis/proxy.py b/lisa/analysis/proxy.py
index a33c8302df609590b88f1f8b86992e9498b99c5c..7b2d380d73abe2e20e4ded764ac9b5e9119ed5b6 100644
--- a/lisa/analysis/proxy.py
+++ b/lisa/analysis/proxy.py
@@ -66,7 +66,7 @@ class AnalysisProxy(Loggable):
         if attr.startswith('__') and attr.endswith('__'):
             return super().__getattribute__(attr)
 
-        logger = self.get_logger()
+        logger = self.logger
 
         # First, try to get the instance of the Analysis that was built if we
         # used it already on that proxy.
diff --git a/lisa/analysis/rta.py b/lisa/analysis/rta.py
index 1191212981693c6a4c5be9d3abfecffcb3d899af..9e103ea08da03def1938b36d8e825ebd1a7f06d2 100644
--- a/lisa/analysis/rta.py
+++ b/lisa/analysis/rta.py
@@ -664,7 +664,7 @@ class RTAEventsAnalysis(TraceAnalysisBase):
         ylabel = 'perf index of "{}"'.format(task)
         series = self.df_rtapp_stats(task)['perf_index']
         mean = series.mean()
-        self.get_logger().info('perf index of task "{}": avg={:.2f} std={:.2f}'
+        self.logger.info('perf index of task "{}": avg={:.2f} std={:.2f}'
                                .format(task, mean, series.std()))
 
         series.hist(bins=bins, ax=axis, alpha=0.4, label=ylabel, figure=axis.get_figure())
@@ -698,7 +698,7 @@ class PerfAnalysis(AnalysisHelpers):
         """
         Load peformance data of an rt-app workload
         """
-        logger = self.get_logger()
+        logger = self.logger
 
         if not task_log_map:
             raise ValueError('No tasks in the task log mapping')
@@ -906,7 +906,7 @@ class PerfAnalysis(AnalysisHelpers):
         ylabel = 'perf index of "{}"'.format(task)
         series = self.get_df(task)['PerfIndex']
         mean = series.mean()
-        self.get_logger().info('perf index of task "{}": avg={:.2f} std={:.2f}'.format(
+        self.logger.info('perf index of task "{}": avg={:.2f} std={:.2f}'.format(
             task, mean, series.std()))
 
         series.hist(bins=bins, ax=axis, alpha=0.4, label=ylabel)
diff --git a/lisa/conf.py b/lisa/conf.py
index 3f08a465451b0be5ca1a86763aee4089ddd8a0d0..bede11eb4403c292e373887c84666cdf75bd07b4 100644
--- a/lisa/conf.py
+++ b/lisa/conf.py
@@ -785,7 +785,7 @@ class MultiSrcConf(MultiSrcConfABC, Loggable, Mapping):
         mapping as a new source.
         """
 
-        logger = self.get_logger()
+        logger = self.logger
         if logger.isEnabledFor(logging.DEBUG):
             caller, filename, lineno = get_call_site(1, exclude_caller_module=True)
             logger.debug('{caller} ({filename}:{lineno}) has set source "{src}":\n{conf}'.format(
@@ -1089,7 +1089,7 @@ class MultiSrcConf(MultiSrcConfABC, Loggable, Mapping):
                 val = self._eval_deferred_val(src, key)
 
 
-        logger = self.get_logger()
+        logger = self.logger
         if not quiet and logger.isEnabledFor(logging.DEBUG):
             caller, filename, lineno = get_call_site(2, exclude_caller_module=True)
             logger.debug('{caller} ({filename}:{lineno}) has used key {key} from source "{src}": {val}'.format(
diff --git a/lisa/energy_meter.py b/lisa/energy_meter.py
index cc8971ca1f6a19f01b98c15c7b3296d43c1dc6fc..2f0a5655f328753d99ec32aff8f7d4efb1fb5243 100644
--- a/lisa/energy_meter.py
+++ b/lisa/energy_meter.py
@@ -144,7 +144,7 @@ class HWMon(EnergyMeter):
 
     def __init__(self, target, channel_map, res_dir=None):
         super().__init__(target, res_dir)
-        logger = self.get_logger()
+        logger = self.logger
 
         # Energy readings
         self.readings = {}
@@ -193,7 +193,7 @@ class HWMon(EnergyMeter):
 
 
     def sample(self):
-        logger = self.get_logger()
+        logger = self.logger
         samples = self._hwmon.take_measurement()
         for s in samples:
             site = s.channel.site
@@ -219,7 +219,7 @@ class HWMon(EnergyMeter):
         for site in self.readings:
             self.readings[site]['delta'] = 0
             self.readings[site]['total'] = 0
-        self.get_logger().debug('RESET: %s', self.readings)
+        self.logger.debug('RESET: %s', self.readings)
 
     def report(self, out_dir, out_file='energy.json'):
         # Retrive energy consumption data
@@ -232,7 +232,7 @@ class HWMon(EnergyMeter):
                                    'Selected channels: {}'.format(
                                        channel, list(nrg.keys())))
             nrg_total = nrg[site]['total']
-            self.get_logger().debug('Energy [%16s]: %.6f', site, nrg_total)
+            self.logger.debug('Energy [%16s]: %.6f', site, nrg_total)
             clusters_nrg[channel] = nrg_total
 
         # Dump data as JSON file
@@ -323,7 +323,7 @@ class AEP(_DevlibContinuousEnergyMeter):
 
     def __init__(self, target, resistor_values, labels=None, device_entry='/dev/ttyACM0', res_dir=None):
         super().__init__(target, res_dir)
-        logger = self.get_logger()
+        logger = self.logger
 
         # Configure channels for energy measurements
         self._instrument = devlib.EnergyProbeInstrument(
@@ -414,7 +414,7 @@ class ACME(EnergyMeter):
           host='baylibre-acme.local', iio_capture_bin='iio-capture',
           res_dir=None):
         super().__init__(target, res_dir)
-        logger = self.get_logger()
+        logger = self.logger
 
         self._iiocapturebin = iio_capture_bin
         self._hostname = host
@@ -453,7 +453,7 @@ class ACME(EnergyMeter):
         Reset energy meter and start sampling from channels specified in the
         target configuration.
         """
-        logger = self.get_logger()
+        logger = self.logger
         # Terminate already running iio-capture instance (if any)
         wait_for_termination = 0
         for proc in psutil.process_iter():
@@ -526,7 +526,7 @@ class ACME(EnergyMeter):
         if delta < self.REPORT_DELAY_S:
             sleep(self.REPORT_DELAY_S - delta)
 
-        logger = self.get_logger()
+        logger = self.logger
         channels_nrg = {}
         channels_stats = {}
         for channel, ch_id in self._channels.items():
diff --git a/lisa/energy_model.py b/lisa/energy_model.py
index c0a34f3a505d6eb8a2e388b14585dce71a904d50..fcb276139ae15966720f1125673c5dab8dceb08e 100644
--- a/lisa/energy_model.py
+++ b/lisa/energy_model.py
@@ -168,7 +168,7 @@ class EnergyModelNode(_CpuTree):
     def __init__(self, active_states, idle_states,
                  cpu=None, children=None, name=None):
         super().__init__(cpu, children)
-        logger = self.get_logger()
+        logger = self.logger
 
         def is_monotonic(l, decreasing=False):
             op = operator.ge if decreasing else operator.le
@@ -343,7 +343,7 @@ class EnergyModel(Serializable, Loggable):
     """
 
     def __init__(self, root_node, root_power_domain, freq_domains):
-        logger = self.get_logger()
+        logger = self.logger
         self.cpus = root_node.cpus
         if self.cpus != tuple(range(len(self.cpus))):
             raise ValueError('CPU IDs [{}] are sparse'.format(self.cpus))
@@ -711,7 +711,7 @@ class EnergyModel(Serializable, Loggable):
 
         num_candidates = len(self.cpus) ** len(tasks)
 
-        logger = self.get_logger()
+        logger = self.logger
         logger.debug(
             '%14s - Searching %d configurations for optimal task placement...',
             'EnergyModel', num_candidates)
diff --git a/lisa/target.py b/lisa/target.py
index 387893efbd3bc39b5e7d20ef51124a7ea9974d17..36a4777eb4dd5455b9d0a210ddefccb0c9c3c846 100644
--- a/lisa/target.py
+++ b/lisa/target.py
@@ -214,7 +214,7 @@ class Target(Loggable, HideExekallID, ExekallTaggable, Configurable):
     ):
 
         super().__init__()
-        logger = self.get_logger()
+        logger = self.logger
 
         self.name = name
 
@@ -496,7 +496,7 @@ class Target(Loggable, HideExekallID, ExekallTaggable, Configurable):
         """
         Initialize the Target
         """
-        logger = self.get_logger()
+        logger = self.logger
         conn_settings = {}
 
         # If the target is Android, we need just (eventually) the device
@@ -647,7 +647,7 @@ class Target(Loggable, HideExekallID, ExekallTaggable, Configurable):
         )
 
     def _get_res_dir(self, root, relative, name, append_time, symlink):
-        logger = self.get_logger()
+        logger = self.logger
         while True:
             time_str = datetime.now().strftime('%Y%m%d_%H%M%S.%f')
             if not name:
@@ -721,7 +721,7 @@ class Target(Loggable, HideExekallID, ExekallTaggable, Configurable):
         """
         Context manager that lets you freeze the userspace
         """
-        logger = self.get_logger()
+        logger = self.logger
         if 'cgroups' not in self.target.modules:
             raise RuntimeError('Could not freeze userspace: "cgroups" devlib module is necessary')
 
@@ -762,7 +762,7 @@ class Target(Loggable, HideExekallID, ExekallTaggable, Configurable):
         # This assumes that freq domains are tied to "idle domains"
         # We'll have to change this if this assumption no longer holds true
 
-        logger = self.get_logger()
+        logger = self.logger
         logger.info('Disabling idle states for all domains')
 
         try:
diff --git a/lisa/tests/base.py b/lisa/tests/base.py
index 41e494c7a4c84709e714a53081d6640dd5d47ba4..686aedc57a1840c8e3cc74df48ae1b3e27058cbe 100644
--- a/lisa/tests/base.py
+++ b/lisa/tests/base.py
@@ -910,7 +910,7 @@ class DmesgTestBundle(TestBundle):
         # Consider as an issue all levels more critical than `level`
         issue_levels = levels[:levels.index(level) + 1]
 
-        logger = self.get_logger()
+        logger = self.logger
 
         if ignored_patterns:
             logger.info('Will ignore patterns in dmesg output: {}'.format(ignored_patterns))
@@ -1147,7 +1147,7 @@ class RTATestBundle(FtraceTestBundle, DmesgTestBundle):
                 if df_filter_task_ids(df, [task_id]).iloc[0].runtime_pct <= threshold
             )
 
-        self.get_logger().info("Ignored PIDs for noise contribution: {}".format(
+        self.logger.info("Ignored PIDs for noise contribution: {}".format(
             ", ".join(map(str, ignored_ids))
         ))
 
diff --git a/lisa/tests/scheduler/load_tracking.py b/lisa/tests/scheduler/load_tracking.py
index 376e17bba3c8d8b5deeee435ddadfdadefcb74f0..e950ba4dced33ed44bfae1cf3a10d8a3242d28cc 100644
--- a/lisa/tests/scheduler/load_tracking.py
+++ b/lisa/tests/scheduler/load_tracking.py
@@ -317,7 +317,7 @@ class InvarianceItem(LoadTrackingBase, ExekallTaggable):
             signal_name == 'load'
             and kernel_version.parts[:2] < (5, 1)
         ):
-            self.get_logger().warning('Load signal is assumed to be CPU invariant, which is true for recent mainline kernels, but may be wrong for {}'.format(
+            self.logger.warning('Load signal is assumed to be CPU invariant, which is true for recent mainline kernels, but may be wrong for {}'.format(
                 kernel_version,
             ))
 
@@ -576,7 +576,7 @@ class Invariance(TestBundle, LoadTrackingHelpers):
         .. seealso:: :class:`InvarianceItem.test_task_util_avg`
         """
 
-        logger = self.get_logger()
+        logger = self.logger
 
         def make_group_bundle(cpu, item_group):
             bundle = AggregatedResultBundle(
@@ -693,7 +693,7 @@ class PELTTask(LoadTrackingBase):
             containing the simulated signal, along with the column of the
             signal as found in the trace.
         """
-        logger = self.get_logger()
+        logger = self.logger
         trace = self.trace
         task = trace.get_task_id(task)
         cpus = trace.analysis.tasks.cpus_of_tasks([task])
diff --git a/lisa/tests/staging/sched_android.py b/lisa/tests/staging/sched_android.py
index 752bb4eb7a053f740771e7c5d901ed3c30c4f4c7..1ea4215eaa08af3cd28381e0693ba91df1ffc339 100644
--- a/lisa/tests/staging/sched_android.py
+++ b/lisa/tests/staging/sched_android.py
@@ -177,7 +177,7 @@ class SchedTuneFreqItem(SchedTuneItemBase):
         """
         kernel_version = self.plat_info['kernel']['version']
         if kernel_version.parts[:2] < (4, 14):
-            self.get_logger().warning('This test requires the RT boost hold, but it may be disabled in {}'.format(kernel_version))
+            self.logger.warning('This test requires the RT boost hold, but it may be disabled in {}'.format(kernel_version))
 
         cpu = self.plat_info['capacity-classes'][-1][0]
         freqs = self.plat_info['freqs'][cpu]
diff --git a/lisa/trace.py b/lisa/trace.py
index 3b952a6cc8f1b17052340db0d2b42022db6f086b..10a2bf384640fb7bd315149bd2282f089f4107b8 100644
--- a/lisa/trace.py
+++ b/lisa/trace.py
@@ -304,7 +304,7 @@ class Trace(Loggable, TraceBase):
 
         super().__init__()
 
-        logger = self.get_logger()
+        logger = self.logger
 
         if plat_info is None:
             plat_info = PlatformInfo()
@@ -338,7 +338,7 @@ class Trace(Loggable, TraceBase):
             max_cpu = max(int(self.df_events(e)['__cpu'].max())
                           for e in self.available_events)
             count = max_cpu + 1
-            self.get_logger().info("Estimated CPU count from trace: %s", count)
+            self.logger.info("Estimated CPU count from trace: %s", count)
             return count
 
     @deprecate('Direct access to underlying ftrace object is discouraged as this is now an implementation detail of that class which could change in the future',
@@ -390,7 +390,7 @@ class Trace(Loggable, TraceBase):
             - SysTrace
         :type trace_format: str
         """
-        logger = self.get_logger()
+        logger = self.logger
         logger.debug('Loading [sched] events from trace in [%s]...', path)
         logger.debug('Parsing events: %s', self.events)
         if trace_format.upper() == 'SYSTRACE' or path.endswith('html'):
@@ -437,7 +437,7 @@ class Trace(Loggable, TraceBase):
         :param key: key to be used for TRAPpy filtering
         :type key: str
         """
-        logger = self.get_logger()
+        logger = self.logger
         available_events = []
         for val in self._ftrace.get_filters(key):
             obj = getattr(self._ftrace, val)
@@ -527,7 +527,7 @@ class Trace(Loggable, TraceBase):
         self.end = self.start + duration
         self.time_range = self.end - self.start
 
-        self.get_logger().debug('Trace contains events from %s to %s',
+        self.logger.debug('Trace contains events from %s to %s',
                                 self.start, self.end)
 
     def get_task_name_pids(self, name, ignore_fork=True):
@@ -837,7 +837,7 @@ class Trace(Loggable, TraceBase):
 
         Also convert between existing field name formats for sched_energy_diff
         """
-        logger = self.get_logger()
+        logger = self.logger
         if not self.has_events('sched_energy_diff') \
            or 'nrg-model' not in self.plat_info:
             return
@@ -918,7 +918,7 @@ class Trace(Loggable, TraceBase):
         """
         Rename some columns and add fake devlib frequency events
         """
-        logger = self.get_logger()
+        logger = self.logger
         if not self.has_events('cpu_frequency_devlib') \
            or 'freq-domains' not in self.plat_info:
             return
diff --git a/lisa/utils.py b/lisa/utils.py
index 06bbbf88f1bda396d1db41808d062e3b48fdef5f..b707177b70da4eeaef509bf380e83f96e892de1d 100644
--- a/lisa/utils.py
+++ b/lisa/utils.py
@@ -101,6 +101,24 @@ class Loggable:
             name += '.' + suffix
         return logging.getLogger(name)
 
+    @property
+    def logger(self):
+        """
+        Default logger, shorthand to ``self.logger``.
+        """
+        try:
+            return self.__dict__['logger']
+        except KeyError:
+            logger = self.get_logger()
+            self.logger = logger
+            return logger
+
+    # Devlib sets logger attribute on some of its classes, which means this
+    # property has to be settable
+    @logger.setter
+    def logger(self, logger):
+        self.__dict__['logger'] = logger
+
     @classmethod
     def log_locals(cls, var_names=None, level='debug'):
         """
diff --git a/lisa/wa_results_collector.py b/lisa/wa_results_collector.py
index 54006ca84f8c4876682400cea8d580968331c2ae..c4807b885c7c558353dcfc810f171b5e90914c8f 100644
--- a/lisa/wa_results_collector.py
+++ b/lisa/wa_results_collector.py
@@ -127,7 +127,7 @@ class WaResultsCollector(Loggable):
                  kernel_repo_path=None, parse_traces=True,
                  use_cached_trace_metrics=True, display_charts=True):
 
-        logger = self.get_logger()
+        logger = self.logger
 
         if base_dir:
             base_dir = os.path.expanduser(base_dir)
@@ -178,7 +178,7 @@ class WaResultsCollector(Loggable):
 
     def _list_wa_dirs(self, base_dir, wa_dirs_re):
         dirs = []
-        logger = self.get_logger()
+        logger = self.logger
         logger.info("Processing WA3 dirs matching [%s], rooted at %s",
                        wa_dirs_re, base_dir)
         wa_dirs_re = re.compile(wa_dirs_re)
@@ -222,7 +222,7 @@ class WaResultsCollector(Loggable):
         # |- pelt-wk1-jankbench-2/
         #      [etc]
 
-        logger = self.get_logger()
+        logger = self.logger
 
         # results.csv contains all the metrics reported by WA for all jobs.
         df = pd.read_csv(os.path.join(wa_dir, 'results.csv'))
@@ -393,7 +393,7 @@ class WaResultsCollector(Loggable):
 
         metric,value,units
         """
-        logger = self.get_logger()
+        logger = self.logger
         cache_path = os.path.join(os.path.dirname(trace_path), 'lisa_trace_metrics.csv')
         if self.use_cached_trace_metrics and os.path.exists(cache_path):
             return pd.read_csv(cache_path)
@@ -489,7 +489,7 @@ class WaResultsCollector(Loggable):
 
         metric,value,units
         """
-        logger = self.get_logger()
+        logger = self.logger
         # return
         # value,metric,units
         extra_metric_list = []
@@ -625,7 +625,7 @@ class WaResultsCollector(Loggable):
         """
         Common helper for getting results to plot for a given metric
         """
-        logger = self.get_logger()
+        logger = self.logger
 
         df = self._select(tag, kernel, test)
         if df.empty:
@@ -931,7 +931,7 @@ class WaResultsCollector(Loggable):
         :param tests: regular expression to filter tests that should be plotted
         :type tests: str
         """
-        logger = self.get_logger()
+        logger = self.logger
 
         if not self.display_charts:
             return
@@ -990,7 +990,7 @@ class WaResultsCollector(Loggable):
         'kernel' column in the results_df uses). If by='tag' then `base_id`
         should be a WA 'tag id' (as named in the WA agenda).
         """
-        logger = self.get_logger()
+        logger = self.logger
         comparisons = []
 
         # I dunno why I wrote this with a namedtuple instead of just a dict or
@@ -1095,7 +1095,7 @@ class WaResultsCollector(Loggable):
         should be a WA 'tag id' (as named in the WA agenda).
         """
 
-        logger = self.get_logger()
+        logger = self.logger
         if not self.display_charts:
             return
 
diff --git a/lisa/wlgen/rta.py b/lisa/wlgen/rta.py
index b3d8fff55cbb3ab3208e70e9ad0e2f6224740b90..cdea99f42c1ac6a29fe01485ac600c8b64fc06e8 100644
--- a/lisa/wlgen/rta.py
+++ b/lisa/wlgen/rta.py
@@ -103,7 +103,7 @@ class RTA(Workload):
 
     def run(self, cpus=None, cgroup=None, background=False, as_root=False):
         super().run(cpus, cgroup, background, as_root)
-        logger = self.get_logger()
+        logger = self.logger
 
         if background:
             # TODO: handle background case
@@ -493,7 +493,7 @@ class Phase(Loggable):
 
         :returns: OrderedDict
         """
-        logger = self.get_logger()
+        logger = self.logger
         phase = OrderedDict()
         # Convert time parameters to integer [us] units
         duration = int(self.duration_s * 1e6)
diff --git a/lisa/wlgen/workload.py b/lisa/wlgen/workload.py
index 5a210b7ada56264258d5213fdae0c2110ee20571..1004d43e7e4673841b5f97f8f15eddf6090a864b 100644
--- a/lisa/wlgen/workload.py
+++ b/lisa/wlgen/workload.py
@@ -93,7 +93,7 @@ class Workload(Loggable):
         self.run_dir = target.execute("mktemp -d -p {} {}".format(
                                       quote(wlgen_dir), quote(temp_fmt))).strip()
 
-        logger = self.get_logger()
+        logger = self.logger
         logger.info("Creating target's run directory: %s", self.run_dir)
 
         res_dir = res_dir if res_dir else target.get_res_dir(
@@ -111,7 +111,7 @@ class Workload(Loggable):
             sessions. For other purposes, use :class:`Workload` instances as a
             context manager.
         """
-        logger = self.get_logger()
+        logger = self.logger
         logger.info("Wiping target run directory: {}".format(self.run_dir))
         self.target.execute("rm -rf {}".format(quote(self.run_dir)))
 
@@ -147,7 +147,7 @@ class Workload(Loggable):
 
         The standard output will be saved into a file in ``self.res_dir``
         """
-        logger = self.get_logger()
+        logger = self.logger
         if not self.command:
             raise RuntimeError("Workload does not specify any command to execute")