Skip to content

gRIBI scale: compact flows to reduce OTG load #5597

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
dracazacula wants to merge 1 commit into
base: main
Choose a base branch
from
+144 −89
Open

gRIBI scale: compact flows to reduce OTG load #5597

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 3 additions & 1 deletion feature/gribi/otg_tests/gribi_full_scale_t1/gribi_full_scale_t1_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -66,6 +66,7 @@ const (

var (
enablePacketCapture = flag.Bool("enable_packet_capture", false, "Enable packet capture and deep packet inspection validation.")
compactOTGFlows = flag.Bool("compact_otg_flows", true, "Compact OTG flows to reduce the number of flows due to OTG port limits.")
)

// ============================================================
Expand Down Expand Up @@ -141,7 +142,8 @@ func TestGRIBIFullScaleT1(t *testing.T) {
} else {
t.Log("Running fixed-size (64B) traffic — all 5 scenarios, 30 Mpps aggregate")
}
cfgplugins.RunEndToEndTrafficValidation(t, ate, dut, ateConfig, tc.UseIMIX, *enablePacketCapture)
cfgplugins.RunEndToEndTrafficValidation(t, ate, dut, ateConfig, tc.UseIMIX, *enablePacketCapture,
*compactOTGFlows)
})
}
}
4 changes: 3 additions & 1 deletion feature/gribi/otg_tests/gribi_full_scale_t2/gribi_full_scale_t2_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -66,6 +66,7 @@ const (

var (
enablePacketCapture = flag.Bool("enable_packet_capture", false, "Enable packet capture and deep packet inspection validation.")
compactOTGFlows = flag.Bool("compact_otg_flows", true, "Compact OTG flows to reduce the number of flows due to OTG port limits.")
)

// ============================================================
Expand Down Expand Up @@ -152,7 +153,8 @@ func TestGRIBIFullScaleT2(t *testing.T) {
} else {
t.Log("Running fixed-size (64B) traffic — all 5 scenarios, 30 Mpps aggregate")
}
cfgplugins.RunEndToEndTrafficValidation(t, ate, dut, ateConfig, tc.UseIMIX, *enablePacketCapture)
cfgplugins.RunEndToEndTrafficValidation(t, ate, dut, ateConfig, tc.UseIMIX, *enablePacketCapture,
*compactOTGFlows)
})
}
}
225 changes: 138 additions & 87 deletions internal/cfgplugins/gribifullscale.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -342,6 +342,16 @@ func EncapDSCPVal(vrfIdx, variant int) uint32 {
return uint32(d2)
}

// getAllEncapDSCPVals returns all DSCP values across all encap VRFs.
func getAllEncapDSCPVals() []uint32 {
dscpVals := make([]uint32, 0, 2*len(encapVRFs))
for vi := range encapVRFs {
d1, d2 := EncapVRFDSCP(vi)
dscpVals = append(dscpVals, uint32(d1), uint32(d2))
}
return dscpVals
}
Comment thread
dracazacula marked this conversation as resolved.

// ConfigureDUT sets up port interfaces, VRFs, and VRF-selection policy.
func ConfigureDUT(t *testing.T, dut *ondatra.DUTDevice) {
t.Helper()
Expand Down Expand Up @@ -972,20 +982,28 @@ func NewIMIXPortFlow(top gosnappi.Config, name string, perGroupPPS uint64) gosna
// Both DSCP values per VRF are now expressed inside a single flow via SetValues to reduce flow count.
//
// encap: 16 VRFs × 2 proto (IPv4+IPv6) = 32
// decap: 16 VRFs = 16
// decap: 16 VRFs (or 1 if compacted) = 16 (or 1)
// reencap: 16 VRFs × 2 src IPs × 2 proto = 64
// transit: 16 VRFs = 16
// repaired: 16 VRFs = 16
// total = 144
// transit: 16 VRFs (or 1 if compacted) = 16 (or 1)
// repaired: 16 VRFs (or 1 if compacted) = 16 (or 1)
// total = 144 (or 99)
//
// Fixed-size: 144 flow groups ✓ (< 256 limit)
// IMIX: 144 flow groups ✓ (SetChoice(WEIGHT) = 1 group per VRF)
func CountBaseFlows() int {
func CountBaseFlows(compactOTGFlows bool) int {
transitFlows := NumEncapVRFs
decapFlows := NumEncapVRFs
repairedFlows := NumEncapVRFs
if compactOTGFlows {
transitFlows = 1
decapFlows = 1
repairedFlows = 1
}
return NumEncapVRFs*2 + // encap IPv4+IPv6
NumEncapVRFs + // decap
decapFlows + // decap
NumEncapVRFs*len(outerSrcs)*2 + // reencap IPv4+IPv6
NumEncapVRFs + // transit
NumEncapVRFs // repaired
transitFlows + // transit
repairedFlows // repaired
}

// MakeFlowCreator returns a function that creates either fixed-size or IMIX flows based on the imix bool.
Expand All @@ -1000,6 +1018,47 @@ func MakeFlowCreator(top gosnappi.Config, pktSize uint32, pps uint64, imix bool)
}
}

// IPIncrement defines the parameters for an incrementing IP pattern in a flow.
type IPIncrement struct {
Start string
Step string
Count uint32
}

// setIPv4Dst applies the given destination IP configuration to the flow pattern.
func setIPv4Dst(dst gosnappi.PatternFlowIpv4Dst, val any) {
switch v := val.(type) {
case string:
dst.SetValue(v)
case []string:
dst.SetValues(v)
case IPIncrement:
dst.Increment().SetStart(v.Start).SetStep(v.Step).SetCount(v.Count)
default:
panic(fmt.Sprintf("unsupported IPv4 destination type: %T", val))
}
}

// createIPv4InIPv4Flow is a helper to reduce duplication when building flows with an IPv4-in-IPv4 header.
func createIPv4InIPv4Flow(newFlow func(string) gosnappi.Flow, name, dstMac, outerSrc string, dscpVals []uint32, outerDst, innerDst any) gosnappi.Flow {
f := newFlow(name)

eth := f.Packet().Add().Ethernet()
eth.Src().SetValue(ATEPort1MAC)
eth.Dst().SetValue(dstMac)

outer := f.Packet().Add().Ipv4()
outer.Src().SetValue(outerSrc)
setIPv4Dst(outer.Dst(), outerDst)
outer.Priority().Dscp().Phb().SetValues(dscpVals)

inner := f.Packet().Add().Ipv4()
inner.Src().SetValue(ATEPort1IPv4)
setIPv4Dst(inner.Dst(), innerDst)

return f
}

// BuildEncapFlows builds fixed-size/imix encap flows for all encap VRFs. IPv4 and IPv6 inners are separate flows since the inner src/dst formats differ.
func BuildEncapFlows(top gosnappi.Config, pktSize uint32, pps uint64, imix bool, dstMac string) []gosnappi.Flow {
flows := make([]gosnappi.Flow, 0)
Expand Down Expand Up @@ -1037,45 +1096,34 @@ func BuildEncapFlows(top gosnappi.Config, pktSize uint32, pps uint64, imix bool,
}

// BuildDecapFlows builds fixed-size/imix decap flows for all encap VRFs. Both DSCPs per VRF are expressed via SetValues in a single flow since the outer header is the same.
func BuildDecapFlows(top gosnappi.Config, pktSize uint32, pps uint64, imix bool, dstMac string) []gosnappi.Flow {
func BuildDecapFlows(top gosnappi.Config, pktSize uint32, pps uint64, imix bool, dstMac string, compact bool) []gosnappi.Flow {
flows := make([]gosnappi.Flow, 0)
decapDsts := ExpandDecapPrefixes()
atePort2Ips, _ := iputil.GenerateIPsWithStep(ATEPort2IPv4Start, NumPort2VLANs, PortIPv4Step)

newFlow := MakeFlowCreator(top, pktSize, pps, imix)

for vi := range encapVRFs {
d1, d2 := EncapVRFDSCP(vi)

f := newFlow(fmt.Sprintf("decap_vrf_%d_src_111", vi))
createFlow := func(name string, dscpVals []uint32, dstIPs []string) {
f := createIPv4InIPv4Flow(newFlow, name, dstMac, IPv4OuterSrc111, dscpVals, decapDsts, dstIPs)

eth := f.Packet().Add().Ethernet()
eth.Src().SetValue(ATEPort1MAC)
eth.Dst().SetValue(dstMac)

outer := f.Packet().Add().Ipv4()
outer.Src().SetValue(IPv4OuterSrc111)
outer.Dst().SetValues(decapDsts)
outer.Priority().Dscp().Phb().SetValues([]uint32{
uint32(d1),
uint32(d2),
})

inner := f.Packet().Add().Ipv4()
inner.Src().SetValue(ATEPort1IPv4)
flows = append(flows, f)
}

// Distribute inner destinations across available ATE IPs to avoid all flows
// stressing the exact same sub-interfaces.
if compact {
createFlow("decap_vrf_all_src_111", getAllEncapDSCPVals(), atePort2Ips)
} else {
subsetCount := max(1, NumPort2VLANs*DecapDestsSubsetPct/100)
var dstIPs []string
for j := 0; j < subsetCount; j++ {
idx := (vi*subsetCount + j) % NumPort2VLANs
dstIPs = append(dstIPs, atePort2Ips[idx])
for vi := range encapVRFs {
d1, d2 := EncapVRFDSCP(vi)
var dstIPs []string
for j := 0; j < subsetCount; j++ {
idx := (vi*subsetCount + j) % NumPort2VLANs
dstIPs = append(dstIPs, atePort2Ips[idx])
}
createFlow(fmt.Sprintf("decap_vrf_%d_src_111", vi), []uint32{uint32(d1), uint32(d2)}, dstIPs)
}
inner.Dst().SetValues(dstIPs)

flows = append(flows, f)
}

return flows
}

Expand Down Expand Up @@ -1132,68 +1180,58 @@ func BuildReencapFlows(top gosnappi.Config, pktSize uint32, pps uint64, imix boo
}

// BuildTransitFlows builds fixed-size/imix transit flows for all encap VRFs.
func BuildTransitFlows(top gosnappi.Config, pktSize uint32, pps uint64, imix bool, dstMac string) []gosnappi.Flow {
func BuildTransitFlows(top gosnappi.Config, pktSize uint32, pps uint64, imix bool, dstMac string, compact bool) []gosnappi.Flow {
flows := make([]gosnappi.Flow, 0)

newFlow := MakeFlowCreator(top, pktSize, pps, imix)

for vi := range encapVRFs {
d1, d2 := EncapVRFDSCP(vi)

f := newFlow(fmt.Sprintf("transit_encap_te_vrf_%d", vi))

eth := f.Packet().Add().Ethernet()
eth.Src().SetValue(ATEPort1MAC)
eth.Dst().SetValue(dstMac)

outer := f.Packet().Add().Ipv4()
outer.Src().SetValue(IPv4OuterSrc111)
outer.Dst().Increment().SetStart(TransitVRF111PrefixStart).SetStep(CommonPrefixStep).SetCount(uint32(NumTransitIPv4))
outer.Priority().Dscp().Phb().SetValues([]uint32{
uint32(d1),
uint32(d2),
})

inner := f.Packet().Add().Ipv4()
inner.Src().SetValue(ATEPort1IPv4)
inner.Dst().SetValue(TransitIPv4InnerDst)
createFlow := func(name string, dscpVals []uint32) {
f := createIPv4InIPv4Flow(newFlow, name, dstMac, IPv4OuterSrc111, dscpVals, IPIncrement{
Start: TransitVRF111PrefixStart,
Step: CommonPrefixStep,
Count: uint32(NumTransitIPv4),
}, TransitIPv4InnerDst)

flows = append(flows, f)
}

if compact {
createFlow("transit_encap_te_vrf_all", getAllEncapDSCPVals())
} else {
for vi := range encapVRFs {
d1, d2 := EncapVRFDSCP(vi)
createFlow(fmt.Sprintf("transit_encap_te_vrf_%d", vi), []uint32{uint32(d1), uint32(d2)})
}
}

return flows
}

// BuildRepairedFlows builds fixed-size/imix flows for all repaired VRFs. One flow per VRF; both DSCPs via SetValues.
func BuildRepairedFlows(top gosnappi.Config, pktSize uint32, pps uint64, imix bool, dstMac string) []gosnappi.Flow {
// BuildRepairedFlows builds fixed-size/imix flows for all repaired VRFs.
func BuildRepairedFlows(top gosnappi.Config, pktSize uint32, pps uint64, imix bool, dstMac string, compact bool) []gosnappi.Flow {
flows := make([]gosnappi.Flow, 0)

newFlow := MakeFlowCreator(top, pktSize, pps, imix)

for vi := range encapVRFs {
d1, d2 := EncapVRFDSCP(vi)

f := newFlow(fmt.Sprintf("repaired_encap_te_vrf_%d", vi))

eth := f.Packet().Add().Ethernet()
eth.Src().SetValue(ATEPort1MAC)
eth.Dst().SetValue(dstMac)

outer := f.Packet().Add().Ipv4()
outer.Src().SetValue(IPv4OuterSrc222)
outer.Dst().Increment().SetStart(TransitVRF222PrefixStart).SetStep(CommonPrefixStep).SetCount(uint32(NumTransitIPv4))
outer.Priority().Dscp().Phb().SetValues([]uint32{
uint32(d1),
uint32(d2),
})

inner := f.Packet().Add().Ipv4()
inner.Src().SetValue(ATEPort1IPv4)
inner.Dst().SetValue(RepairIPv4InnerDst)
createFlow := func(name string, dscpVals []uint32) {
f := createIPv4InIPv4Flow(newFlow, name, dstMac, IPv4OuterSrc222, dscpVals, IPIncrement{
Start: TransitVRF222PrefixStart,
Step: CommonPrefixStep,
Count: uint32(NumTransitIPv4),
}, RepairIPv4InnerDst)

flows = append(flows, f)
}

if compact {
createFlow("repaired_encap_te_vrf_all", getAllEncapDSCPVals())
} else {
for vi := range encapVRFs {
d1, d2 := EncapVRFDSCP(vi)
createFlow(fmt.Sprintf("repaired_encap_te_vrf_%d", vi), []uint32{uint32(d1), uint32(d2)})
}
}

return flows
}

Expand Down Expand Up @@ -1223,9 +1261,9 @@ func GetDUTMACAddress(t *testing.T, ate *ondatra.ATEDevice, intfName string, nei
}

// RunEndToEndTrafficValidation executes the end-to-end traffic validation for all scenarios. It registers flows, configures capture, runs traffic, and validates via otgvalidationhelpers and packetvalidationhelpers.
func RunEndToEndTrafficValidation(t *testing.T, ate *ondatra.ATEDevice, dut *ondatra.DUTDevice, top gosnappi.Config, imix bool, enablePacketCapture bool) {
func RunEndToEndTrafficValidation(t *testing.T, ate *ondatra.ATEDevice, dut *ondatra.DUTDevice, top gosnappi.Config, imix bool, enablePacketCapture bool, compactOTGFlows bool) {
Comment thread
dracazacula marked this conversation as resolved.
t.Helper()
baseFlows := CountBaseFlows()
baseFlows := CountBaseFlows(compactOTGFlows)
perFlowPPS := TrafficRateMpps / uint64(baseFlows)
if perFlowPPS == 0 {
perFlowPPS = 1
Expand All @@ -1243,13 +1281,20 @@ func RunEndToEndTrafficValidation(t *testing.T, ate *ondatra.ATEDevice, dut *ond
// from the flow's position in the slice so both DSCP values for that VRF
// can be stored in ExpectedDSCPs — either is a valid egress DSCP.
addFlows := func(builtFlows []gosnappi.Flow, sc TrafficScenario, outerSrc string, wantEncap bool, flowsPerVRF int) {
isCompacted := compactOTGFlows && (sc == ScenarioDecap || sc == ScenarioTransit || sc == ScenarioRepaired)
for fi, f := range builtFlows {
vi := fi / flowsPerVRF // VRF index from position in scenario slice
d1, d2 := EncapVRFDSCP(vi)
var expectedDSCPs []uint32
if isCompacted {
expectedDSCPs = getAllEncapDSCPVals()
} else {
vi := fi / flowsPerVRF // VRF index from position in scenario slice
d1, d2 := EncapVRFDSCP(vi)
expectedDSCPs = []uint32{uint32(d1), uint32(d2)}
}
expectations[f.Name()] = FlowExpectation{
Scenario: sc,
ExpectedOuterSrc: outerSrc,
ExpectedDSCPs: []uint32{uint32(d1), uint32(d2)},
ExpectedDSCPs: expectedDSCPs,
WantEncapPresent: wantEncap,
DecapPrefixSet: decapPfxSet,
}
Expand All @@ -1273,10 +1318,16 @@ func RunEndToEndTrafficValidation(t *testing.T, ate *ondatra.ATEDevice, dut *ond
multiplier int
}{
{BuildEncapFlows, ScenarioEncap, IPv4OuterSrc111, true, 2},
{BuildDecapFlows, ScenarioDecap, "", false, 1},
{func(top gosnappi.Config, pktSize uint32, pps uint64, imix bool, dstMac string) []gosnappi.Flow {
return BuildDecapFlows(top, pktSize, pps, imix, dstMac, compactOTGFlows)
}, ScenarioDecap, "", false, 1},
Comment thread
dracazacula marked this conversation as resolved.
{BuildReencapFlows, ScenarioReencap, "", true, 4},
{BuildTransitFlows, ScenarioTransit, IPv4OuterSrc111, true, 1},
{BuildRepairedFlows, ScenarioRepaired, IPv4OuterSrc222, true, 1},
{func(top gosnappi.Config, pktSize uint32, pps uint64, imix bool, dstMac string) []gosnappi.Flow {
return BuildTransitFlows(top, pktSize, pps, imix, dstMac, compactOTGFlows)
}, ScenarioTransit, IPv4OuterSrc111, true, 1},
{func(top gosnappi.Config, pktSize uint32, pps uint64, imix bool, dstMac string) []gosnappi.Flow {
return BuildRepairedFlows(top, pktSize, pps, imix, dstMac, compactOTGFlows)
}, ScenarioRepaired, IPv4OuterSrc222, true, 1},
}

// Fetch MAC address for port1.
Expand Down
Loading