diff --git a/README.md b/README.md
index 96d8750..4816d5b 100644
--- a/README.md
+++ b/README.md
@@ -54,11 +54,15 @@ most of the moving parts involved:
 ## Design Goals
 
 * Protect all devices in a network with zero client-side configuration or software
-* Dynamically throttle traffic from clients to configurable sites or groups of sites to "train" good habits [1]
+* Dynamically throttle traffic from clients to configurable sites or groups of sites to "train" good habits
 * Use only IP and (sniffed) DNS (i.e. as close to a pure L4 solution as possible)
 * Introduce no latency on "good" traffic
 * Usable on minimal hardware like an Orange Pi R1plus or similar
 
+## Current milestones 
+
+Milestone 8: Improved overall installation experience and support
+for lower-end 32-bit arm devices such as Orange Pi Zero/R1.
 
 ## References & Acknowlegements
 
diff --git a/config/config.go b/config/config.go
index 460c2fa..906a653 100644
--- a/config/config.go
+++ b/config/config.go
@@ -3,6 +3,7 @@ package config
 import (
 	"log"
 	"os"
+	"time"
 
 	"gopkg.in/yaml.v3"
 )
@@ -16,6 +17,9 @@ var baseConfig = Config{
 	DnsDb:         "/root/dns.db",
 	PrometheusUrl: "http://localhost:9090",
 
+	MapCollectionInterval: time.Second * 30,
+	PolicyCheckInterval:   time.Minute * 1,
+
 	// Start degrading from 1mbit down to 10kbits, over 5 bandwidth classes
 	NumQdiscClasses: 5,
 	StartRateKbs:    1000,
@@ -23,7 +27,7 @@ var baseConfig = Config{
 	MinRateKbs:      50,
 
 	ActivityThresholdBytes: 100,
-	PolicyBackoffMinutes:   3,
+	PolicyBackoffInterval:  time.Minute * 3,
 }
 
 var Cfg = baseConfig
@@ -39,6 +43,11 @@ type Config struct {
 	DnsDb         string `yaml:"dns_db"`
 	PrometheusUrl string `yaml:"prometheus_url"`
 
+	// Operational parameters for tuning use on SBCs with different
+	// capabilities
+	MapCollectionInterval time.Duration `yaml:"map_collection_interval"`
+	PolicyCheckInterval   time.Duration `yaml:"policy_check_interval"`
+
 	// Parameters for the htb / netem qdiscs and classes we'll create on the fly
 	// WanMbs represents the max bandwidth of your downstream internet connection
 	// The MinRateKbs and MaxDelayMs represent the maximum bandwidth degradation ;
@@ -58,10 +67,10 @@ type Config struct {
 	// the number of bytes considered "active" for the prom rate
 	// queries
 	ActivityThresholdBytes int `yaml:"activity_threshold_bytes"`
-	// How many minutes before we'll consider a change to policy;
+	// How long before we'll consider a change to policy;
 	// setting this lower will cause clients to be moved up and
 	// down bandwidth classes faster
-	PolicyBackoffMinutes int `yaml:"policy_backoff_minutes"`
+	PolicyBackoffInterval time.Duration `yaml:"policy_backoff_interval"`
 }
 
 func ParseConfig(file string) {
diff --git a/config/nethadone.yml b/config/nethadone.yml
index ace036d..8ed723e 100644
--- a/config/nethadone.yml
+++ b/config/nethadone.yml
@@ -7,6 +7,18 @@ cfg_db: "/root/cfg.db"
 dns_db: "/root/dns.db"
 prometheus_url: "http://localhost:9090"
 
+###
+# Operational parameters
+
+# How frequently we parse the eBPF-level maps for ip bandwidth usage
+# For sites with many clients, and on lower-end hardware such as 
+# 32-bit orange pi zero/r1, this shouldn't be more frequent than 
+# 30s until the parsing code can be optimized
+map_collection_interval: 30s
+
+# How frequently we check for policy violations; 
+policy_check_interval: 1m
+
 # Parameters for the htb # netem qdiscs and classes we'll create on the fly
 # WanMbs represents the max bandwidth of your downstream internet connection
 # The MinRateKbs and MaxDelayMs represent the maximum bandwidth degradation ;
@@ -30,7 +42,7 @@ min_rate_kbs: 50
 # this can be set to a very low number like 0. For more bandwidth heavy
 # sites, a larger number may make sense. 
 activity_threshold_bytes: 1
-# How many minutes before we'll consider a change to policy;
+# How long before we'll consider a change to policy;
 # setting this lower will cause clients to be moved up and
 # down bandwidth classes faster
-policy_backoff_minutes: 2
\ No newline at end of file
+policy_backoff_interval: 2m
\ No newline at end of file
diff --git a/database/prom.go b/database/prom.go
index b4e9d96..112692f 100644
--- a/database/prom.go
+++ b/database/prom.go
@@ -30,9 +30,10 @@ func GetSrcGlobUsage(rate int, mins int, k int, above bool) model.Vector {
 	tm := time.Now().Add(-1 * dr)
 	result, warnings, err := queryAPI.Query(context.Background(), pql, tm)
 	if err != nil {
-		log.Println("error trying to query prometheus: ", err, warnings)
-	}
-	sample := result.(model.Vector)
+		log.Println("unable to query prometheus, policy changes will not be possible: ", err, warnings)
+		return model.Vector{}
+	} else {
 
-	return sample
+		return result.(model.Vector)
+	}
 }
diff --git a/handlers/prom.go b/handlers/prom.go
index 3eef5ae..ab6df56 100644
--- a/handlers/prom.go
+++ b/handlers/prom.go
@@ -6,6 +6,7 @@ import (
 	"time"
 
 	"github.com/VictoriaMetrics/metrics"
+	"github.com/atomic77/nethadone/config"
 )
 
 func InitMetrics() {
@@ -15,12 +16,12 @@ func InitMetrics() {
 }
 
 func pollMetrics() {
-	log.Println("Setting up metrics collector")
-	for range time.Tick(time.Second * 15) {
+	log.Println("Setting up metrics collector on ", config.Cfg.MapCollectionInterval, " interval")
+	for range time.Tick(config.Cfg.MapCollectionInterval) {
 		// TODO Once we get into the 1000s of IP pairs being tracked, the
 		// performance of this method becomes quite slow
 		bl := getBandwidthList(false)
-		log.Println("Tick happened, collected ", len(bl), " pairs")
+		log.Println("Collected bandwidth stats from ", len(bl), " ip pairs")
 		for _, b := range bl {
 			s := fmt.Sprintf(
 				`ip_pair_vic_bytes_total{%s="%s", %s="%s", %s="%s", %s="%s"}`,
diff --git a/policy/simpleloadavg.go b/policy/simpleloadavg.go
index fe4d84c..56912f4 100644
--- a/policy/simpleloadavg.go
+++ b/policy/simpleloadavg.go
@@ -74,7 +74,7 @@ func increaseThrottling() bool {
 			continue
 		}
 
-		backOff := time.Now().Add(time.Minute * time.Duration(config.Cfg.PolicyBackoffMinutes))
+		backOff := time.Now().Add(config.Cfg.PolicyBackoffInterval)
 		if p.Tstamp.After(backOff) {
 			log.Println("too soon to update policy for ", src_ip, glob)
 		} else if p.Class > config.Cfg.NumQdiscClasses*10 {
@@ -109,7 +109,7 @@ func decreasingThrottling() bool {
 
 		// After testing set this back to 5 or some more sensible default
 		// from configuration
-		backOff := time.Now().Add(time.Minute * -1)
+		backOff := time.Now().Add(config.Cfg.PolicyBackoffInterval)
 		if p.Tstamp.After(backOff) {
 			log.Println("too soon to update policy for ", src_ip, glob)
 		} else if p.Class <= 10 {
@@ -126,8 +126,8 @@ func decreasingThrottling() bool {
 }
 
 func pollPolicyCheck() {
-	log.Println("Setting up metrics collector")
-	for range time.Tick(time.Minute * 1) {
+	log.Println("Setting up policy checker on ", config.Cfg.PolicyCheckInterval, " interval ")
+	for range time.Tick(config.Cfg.PolicyCheckInterval) {
 		log.Println("Checking policy ")
 		changed := increaseThrottling()
 		changed = changed || decreasingThrottling()
@@ -153,8 +153,3 @@ func applyPolicies() {
 	handlers.ApplyPolicies(&allPolicies)
 
 }
-func pollPolicyApply() {
-	for range time.Tick(time.Minute * 1) {
-		applyPolicies()
-	}
-}