func BytesPerMinutePipeline(levelDbManager store.Manager, bytesPerHourPostgresStore store.Writer) transformer.Pipeline {
tracesStore := levelDbManager.Seeker("traces")
mappedStore := levelDbManager.ReadingWriter("bytesperminute-mapped")
bytesPerMinuteStore := levelDbManager.ReadingWriter("bytesperminute")
bytesPerHourStore := levelDbManager.ReadingWriter("bytesperhour")
traceKeyRangesStore := levelDbManager.ReadingDeleter("bytesperminute-trace-key-ranges")
consolidatedTraceKeyRangesStore := levelDbManager.ReadingDeleter("bytesperminute-consolidated-trace-key-ranges")
return append([]transformer.PipelineStage{
transformer.PipelineStage{
Name: "BytesPerMinuteMapper",
Reader: store.NewRangeExcludingReader(tracesStore, traceKeyRangesStore),
Transformer: transformer.MakeDoTransformer(bytesPerMinuteMapper(transformer.NewNonce())),
Writer: mappedStore,
},
transformer.PipelineStage{
Name: "BytesPerMinuteReducer",
Reader: mappedStore,
Transformer: transformer.TransformFunc(bytesPerMinuteReducer),
Writer: bytesPerMinuteStore,
},
transformer.PipelineStage{
Name: "BytesPerHourReducer",
Reader: bytesPerMinuteStore,
Transformer: transformer.TransformFunc(bytesPerHourReducer),
Writer: bytesPerHourStore,
},
transformer.PipelineStage{
Name: "BytesPerHourPostgres",
Reader: bytesPerHourStore,
Writer: bytesPerHourPostgresStore,
},
}, TraceKeyRangesPipeline(store.NewRangeExcludingReader(tracesStore, traceKeyRangesStore), traceKeyRangesStore, consolidatedTraceKeyRangesStore)...)
}
func BytesPerDevicePipeline(levelDbManager store.Manager, bytesPerDevicePostgresStore store.Writer) transformer.Pipeline {
tracesStore := levelDbManager.Seeker("traces")
availabilityIntervalsStore := levelDbManager.Seeker("consistent-ranges")
sessionsStore := levelDbManager.ReadingDeleter("bytesperdevice-session")
addressTableStore := levelDbManager.SeekingWriter("bytesperdevice-address-table")
flowTableStore := levelDbManager.SeekingWriter("bytesperdevice-flow-table")
packetsStore := levelDbManager.SeekingWriter("bytesperdevice-packets")
flowIdToMacStore := levelDbManager.SeekingWriter("bytesperdevice-flow-id-to-mac")
flowIdToMacsStore := levelDbManager.SeekingWriter("bytesperdevice-flow-id-to-macs")
bytesPerDeviceUnreducedStore := levelDbManager.SeekingWriter("bytesperdevice-unreduced")
bytesPerDeviceSessionStore := levelDbManager.ReadingWriter("bytesperdevice-reduced-sessions")
bytesPerDeviceStore := levelDbManager.ReadingWriter("bytesperdevice")
traceKeyRangesStore := levelDbManager.ReadingDeleter("bytesperdevice-trace-key-ranges")
consolidatedTraceKeyRangesStore := levelDbManager.ReadingDeleter("bytesperdevice-consolidated-trace-key-ranges")
newTracesStore := store.NewRangeExcludingReader(store.NewRangeIncludingReader(tracesStore, availabilityIntervalsStore), traceKeyRangesStore)
return append([]transformer.PipelineStage{
transformer.PipelineStage{
Name: "BytesPerDeviceMapper",
Reader: newTracesStore,
Transformer: transformer.MakeMultipleOutputsDoFunc(bytesPerDeviceMapper, 3),
Writer: store.NewMuxingWriter(addressTableStore, flowTableStore, packetsStore),
},
SessionPipelineStage(newTracesStore, sessionsStore),
transformer.PipelineStage{
Name: "JoinMacAndFlowId",
Reader: store.NewPrefixIncludingReader(store.NewDemuxingSeeker(addressTableStore, flowTableStore), sessionsStore),
Transformer: transformer.TransformFunc(joinMacAndFlowId),
Writer: flowIdToMacStore,
},
transformer.PipelineStage{
Name: "FlattenMacAddresses",
Reader: store.NewPrefixIncludingReader(flowIdToMacStore, sessionsStore),
Transformer: transformer.TransformFunc(flattenMacAddresses),
Writer: flowIdToMacsStore,
},
transformer.PipelineStage{
Name: "JoinMacAndSizes",
Reader: store.NewPrefixIncludingReader(store.NewDemuxingSeeker(flowIdToMacsStore, packetsStore), sessionsStore),
Transformer: transformer.TransformFunc(joinMacAndSizes),
Writer: bytesPerDeviceUnreducedStore,
},
transformer.PipelineStage{
Name: "ReduceBytesPerDeviceSession",
Reader: store.NewPrefixIncludingReader(bytesPerDeviceUnreducedStore, sessionsStore),
Transformer: transformer.TransformFunc(reduceBytesPerDeviceSession),
Writer: bytesPerDeviceSessionStore,
},
transformer.PipelineStage{
Name: "ReduceBytesPerDevice",
Reader: bytesPerDeviceSessionStore,
Transformer: transformer.TransformFunc(reduceBytesPerDevice),
Writer: bytesPerDeviceStore,
},
transformer.PipelineStage{
Name: "BytesPerDevicePostgres",
Reader: bytesPerDeviceStore,
Writer: bytesPerDevicePostgresStore,
},
}, TraceKeyRangesPipeline(newTracesStore, traceKeyRangesStore, consolidatedTraceKeyRangesStore)...)
}
func AvailabilityPipeline(levelDbManager store.Manager, jsonWriter io.Writer, timestamp int64) transformer.Pipeline {
tracesStore := levelDbManager.Seeker("traces")
intervalsStore := levelDbManager.ReadingWriter("availability-intervals")
consolidatedStore := levelDbManager.ReadingDeleter("availability-consolidated")
nodesStore := levelDbManager.ReadingDeleter("availability-nodes")
excludeRangesStore := levelDbManager.ReadingDeleter("availability-done")
consistentRangesStore := levelDbManager.ReadingDeleter("consistent-ranges")
return []transformer.PipelineStage{
transformer.PipelineStage{
Name: "AvailabilityIntervals",
Reader: store.NewRangeExcludingReader(tracesStore, excludeRangesStore),
Transformer: transformer.TransformFunc(availabilityIntervals),
Writer: intervalsStore,
},
transformer.PipelineStage{
Name: "ConsolidateAvailabilityIntervals",
Reader: intervalsStore,
Transformer: transformer.TransformFunc(consolidateAvailabilityIntervals),
Writer: store.NewTruncatingWriter(consolidatedStore),
},
transformer.PipelineStage{
Name: "AvailabilityReducer",
Reader: consolidatedStore,
Transformer: transformer.TransformFunc(availabilityReducer),
Writer: store.NewTruncatingWriter(nodesStore),
},
transformer.PipelineStage{
Name: "AvailabilityJson",
Reader: nodesStore,
Writer: &availabilityJsonStore{writer: jsonWriter, timestamp: timestamp},
},
transformer.PipelineStage{
Name: "GenerateExcludedRanges",
Reader: consolidatedStore,
Transformer: transformer.MakeMapFunc(generateExcludedRanges),
Writer: store.NewTruncatingWriter(excludeRangesStore),
},
transformer.PipelineStage{
Name: "GenerateConsistentRanges",
Reader: excludeRangesStore,
Transformer: transformer.MakeDoFunc(generateConsistentRanges),
Writer: store.NewTruncatingWriter(consistentRangesStore),
},
}
}
func AggregateStatisticsPipeline(levelDbManager store.Manager, jsonWriter io.Writer) transformer.Pipeline {
tracesStore := levelDbManager.Seeker("traces")
availabilityIntervalsStore := levelDbManager.Seeker("consistent-ranges")
traceAggregatesStore := levelDbManager.SeekingWriter("statistics-trace-aggregates")
sessionAggregatesStore := levelDbManager.ReadingWriter("statistics-session-aggregates")
nodeAggregatesStore := levelDbManager.ReadingWriter("statistics-node-aggregates")
sessionsStore := levelDbManager.ReadingDeleter("statistics-sessions")
traceKeyRangesStore := levelDbManager.ReadingDeleter("statistics-trace-key-ranges")
consolidatedTraceKeyRangesStore := levelDbManager.ReadingDeleter("statistics-consolidated-trace-key-ranges")
newTracesStore := store.NewRangeExcludingReader(store.NewRangeIncludingReader(tracesStore, availabilityIntervalsStore), traceKeyRangesStore)
return append([]transformer.PipelineStage{
transformer.PipelineStage{
Name: "AggregateStatisticsMapper",
Reader: store.NewRangeExcludingReader(tracesStore, traceKeyRangesStore),
Transformer: transformer.MakeMapFunc(aggregateStatisticsMapper),
Writer: traceAggregatesStore,
},
SessionPipelineStage(newTracesStore, sessionsStore),
transformer.PipelineStage{
Name: "AggregateStatisticsReduceBySession",
Reader: store.NewPrefixIncludingReader(traceAggregatesStore, sessionsStore),
Transformer: transformer.TransformFunc(aggregateStatisticsReduceBySession),
Writer: sessionAggregatesStore,
},
transformer.PipelineStage{
Name: "AggregateStatisticsReducer",
Reader: sessionAggregatesStore,
Transformer: transformer.TransformFunc(aggregateStatisticsReducer),
Writer: nodeAggregatesStore,
},
transformer.PipelineStage{
Name: "AggregateStatisticsJson",
Reader: nodeAggregatesStore,
Writer: &aggregateStatisticsJsonStore{writer: jsonWriter},
},
}, TraceKeyRangesPipeline(newTracesStore, traceKeyRangesStore, consolidatedTraceKeyRangesStore)...)
}
func SummarizePipeline(levelDbManager store.Manager, csvManager store.Manager) transformer.Pipeline {
statsStore := levelDbManager.Reader("stats")
statsWithHourStore := levelDbManager.ReadingDeleter("stats-with-hour")
statsWithDayStore := levelDbManager.ReadingDeleter("stats-with-day")
statsWithReceivedTimestampStore := levelDbManager.ReadingDeleter("stats-with-received-timestamp")
interarrivalTimesStore := levelDbManager.ReadingDeleter("interarrival-times")
sizeSummaryStore := levelDbManager.ReadingWriter("size-summary")
sizeSummaryByHourStore := levelDbManager.ReadingWriter("size-summary-by-hour")
sizeSummaryByDayStore := levelDbManager.ReadingWriter("size-summary-by-day")
interarrivalTimesSummaryStore := levelDbManager.ReadingWriter("interarrival-times-summary")
sizePerDayStore := levelDbManager.ReadingWriter("sizes-by-day")
sizeSummaryWriter := makeSummaryCsvWriter(csvManager, "size-summary.csv")
sizeSummaryByHourWriter := makeSummaryByTimestampCsvWriter(csvManager, "size-summary-by-hour.csv")
sizeSummaryByDayWriter := makeSummaryByTimestampCsvWriter(csvManager, "size-summary-by-day.csv")
interarrivalTimesSummaryWriter := makeSummaryCsvWriter(csvManager, "interarrival-times-summary.csv")
sizesPerDayWriter := csvManager.Writer("sizes-per-day.csv", []string{"experiment", "node", "timestamp"}, []string{"count"}, new(string), new(string), new(int64), new(int64))
return []transformer.PipelineStage{
transformer.PipelineStage{
Name: "SummarizeSizes",
Reader: statsStore,
Transformer: transformer.TransformFunc(summarizeSizes),
Writer: sizeSummaryStore,
},
transformer.PipelineStage{
Name: "RekeyStatsByHour",
Reader: statsStore,
Transformer: transformer.MakeMapFunc(rekeyStatsByHour),
Writer: store.NewTruncatingWriter(statsWithHourStore),
},
transformer.PipelineStage{
Name: "SummarizeSizesByHour",
Reader: statsWithHourStore,
Transformer: transformer.TransformFunc(summarizeSizesByTimestamp),
Writer: sizeSummaryByHourStore,
},
transformer.PipelineStage{
Name: "RekeyStatsByDay",
Reader: statsStore,
Transformer: transformer.MakeMapFunc(rekeyStatsByDay),
Writer: store.NewTruncatingWriter(statsWithDayStore),
},
transformer.PipelineStage{
Name: "SummarizeSizesByDay",
Reader: statsWithDayStore,
Transformer: transformer.TransformFunc(summarizeSizesByTimestamp),
Writer: sizeSummaryByDayStore,
},
transformer.PipelineStage{
Name: "RekeyStatsByReceivedTimestamp",
Reader: statsStore,
Transformer: transformer.MakeMapFunc(rekeyStatsByReceviedTimestamp),
Writer: store.NewTruncatingWriter(statsWithReceivedTimestampStore),
},
transformer.PipelineStage{
Name: "ComputeInterarrivalTimes",
Reader: statsWithReceivedTimestampStore,
Transformer: transformer.TransformFunc(computeInterarrivalTimes),
Writer: store.NewTruncatingWriter(interarrivalTimesStore),
},
transformer.PipelineStage{
Name: "SummarizeInterarrival",
Reader: interarrivalTimesStore,
Transformer: transformer.TransformFunc(summarizeInterarrivalTimes),
Writer: interarrivalTimesSummaryStore,
},
transformer.PipelineStage{
Name: "SummarizeSizesPerDay",
Reader: statsStore,
Transformer: transformer.TransformFunc(summarizeSizesPerDay),
Writer: sizePerDayStore,
},
transformer.PipelineStage{
Name: "AggregateExperimentsPerDay",
Reader: sizePerDayStore,
Transformer: transformer.TransformFunc(aggregateSizesPerDay),
Writer: sizePerDayStore,
},
transformer.PipelineStage{
Name: "WriteSizesSummary",
Reader: sizeSummaryStore,
Writer: sizeSummaryWriter,
},
transformer.PipelineStage{
Name: "WriteSizesSummaryByHour",
Reader: sizeSummaryByHourStore,
Writer: sizeSummaryByHourWriter,
},
transformer.PipelineStage{
Name: "WriteSizesSummaryByDay",
Reader: sizeSummaryByDayStore,
Writer: sizeSummaryByDayWriter,
},
transformer.PipelineStage{
Name: "WriteInterarrivalTimesSummary",
Reader: interarrivalTimesSummaryStore,
Writer: interarrivalTimesSummaryWriter,
},
transformer.PipelineStage{
Name: "WriteSizePerDaySummary",
Reader: sizePerDayStore,
Writer: sizesPerDayWriter,
},
}
}
func BytesPerDomainPipeline(levelDbManager store.Manager, bytesPerDomainPostgresStore store.Writer) transformer.Pipeline {
tracesStore := levelDbManager.Seeker("traces")
availabilityIntervalsStore := levelDbManager.Seeker("consistent-ranges")
traceKeyRangesStore := levelDbManager.ReadingDeleter("bytesperdomain-trace-key-ranges")
consolidatedTraceKeyRangesStore := levelDbManager.ReadingDeleter("bytesperdomain-consolidated-trace-key-ranges")
addressIdTableStore := levelDbManager.SeekingWriter("bytesperdomain-address-id-table")
aRecordTableStore := levelDbManager.SeekingWriter("bytesperdomain-a-record-table")
cnameRecordTableStore := levelDbManager.SeekingWriter("bytesperdomain-cname-record-table")
flowIpsTableStore := levelDbManager.SeekingWriter("bytesperdomain-flow-ips-table")
addressIpTableStore := levelDbManager.SeekingWriter("bytesperdomain-address-ip-table")
bytesPerTimestampShardedStore := levelDbManager.SeekingWriter("bytesperdomain-bytes-per-timestamp-sharded")
whitelistStore := levelDbManager.SeekingWriter("bytesperdomain-whitelist")
aRecordsWithMacStore := levelDbManager.SeekingWriter("bytesperdomain-a-records-with-mac")
cnameRecordsWithMacStore := levelDbManager.SeekingWriter("bytesperdomain-cname-records-with-mac")
allDnsMappingsStore := levelDbManager.SeekingWriter("bytesperdomain-all-dns-mappings")
allWhitelistedMappingsStore := levelDbManager.SeekingWriter("bytesperdomain-all-whitelisted-mappings")
flowMacsTableStore := levelDbManager.SeekingWriter("bytesperdomain-flow-macs-table")
flowDomainsTableStore := levelDbManager.SeekingWriter("bytesperdomain-flow-domains-table")
flowDomainsGroupedTableStore := levelDbManager.SeekingWriter("bytesperdomain-flow-domains-grouped-table")
bytesPerDomainShardedStore := levelDbManager.ReadingWriter("bytesperdomain-bytes-per-domain-sharded")
bytesPerDomainPerDeviceStore := levelDbManager.ReadingWriter("bytesperdomain-bytes-per-domain-per-device")
bytesPerDomainStore := levelDbManager.ReadingWriter("bytesperdomain-bytes-per-domain")
sessionsStore := levelDbManager.ReadingDeleter("bytesperdomain-sessions")
excludeOldSessions := func(stor store.Seeker) store.Seeker {
return store.NewPrefixIncludingReader(stor, sessionsStore)
}
newTracesStore := store.NewRangeExcludingReader(store.NewRangeIncludingReader(tracesStore, availabilityIntervalsStore), traceKeyRangesStore)
return append([]transformer.PipelineStage{
transformer.PipelineStage{
Name: "BytesPerDomainMapper",
Reader: newTracesStore,
Transformer: transformer.MakeMultipleOutputsDoFunc(bytesPerDomainMapper, 7),
Writer: store.NewMuxingWriter(addressIdTableStore, aRecordTableStore, cnameRecordTableStore, flowIpsTableStore, addressIpTableStore, bytesPerTimestampShardedStore, whitelistStore),
},
SessionPipelineStage(newTracesStore, sessionsStore),
transformer.PipelineStage{
Name: "JoinAAddressIdsWithMacAddresses",
Reader: excludeOldSessions(store.NewDemuxingSeeker(addressIdTableStore, aRecordTableStore)),
Transformer: transformer.TransformFunc(joinAddressIdsWithMacAddresses),
Writer: aRecordsWithMacStore,
},
transformer.PipelineStage{
Name: "JoinCnameAddressIdsWithMacAddresses",
Reader: excludeOldSessions(store.NewDemuxingSeeker(addressIdTableStore, cnameRecordTableStore)),
Transformer: transformer.TransformFunc(joinAddressIdsWithMacAddresses),
Writer: cnameRecordsWithMacStore,
},
transformer.PipelineStage{
Name: "JoinARecordsWithCnameRecords",
Reader: excludeOldSessions(store.NewDemuxingSeeker(aRecordsWithMacStore, cnameRecordsWithMacStore)),
Transformer: transformer.TransformFunc(joinARecordsWithCnameRecords),
Writer: allDnsMappingsStore,
},
transformer.PipelineStage{
Name: "EmitARecords",
Reader: excludeOldSessions(aRecordsWithMacStore),
Transformer: transformer.MakeDoFunc(emitARecords),
Writer: allDnsMappingsStore,
},
transformer.PipelineStage{
Name: "JoinDomainsWithWhitelist",
Reader: excludeOldSessions(store.NewDemuxingSeeker(whitelistStore, allDnsMappingsStore)),
Transformer: transformer.TransformFunc(joinDomainsWithWhitelist),
Writer: allWhitelistedMappingsStore,
},
transformer.PipelineStage{
Name: "JoinMacWithFlowId",
Reader: excludeOldSessions(store.NewDemuxingSeeker(addressIpTableStore, flowIpsTableStore)),
Transformer: transformer.TransformFunc(joinMacWithFlowId),
Writer: flowMacsTableStore,
},
transformer.PipelineStage{
Name: "JoinWhitelistedDomainsWithFlows",
Reader: excludeOldSessions(store.NewDemuxingSeeker(allWhitelistedMappingsStore, flowMacsTableStore)),
Transformer: transformer.TransformFunc(joinWhitelistedDomainsWithFlows),
Writer: flowDomainsTableStore,
},
transformer.PipelineStage{
Name: "GroupDomainsAndMacAddresses",
Reader: excludeOldSessions(flowDomainsTableStore),
Transformer: transformer.TransformFunc(groupDomainsAndMacAddresses),
Writer: flowDomainsGroupedTableStore,
},
transformer.PipelineStage{
Name: "JoinDomainsWithSizes",
Reader: excludeOldSessions(store.NewDemuxingSeeker(flowDomainsGroupedTableStore, bytesPerTimestampShardedStore)),
Transformer: transformer.TransformFunc(joinDomainsWithSizes),
Writer: bytesPerDomainShardedStore,
},
transformer.PipelineStage{
Name: "FlattenIntoBytesPerDevice",
Reader: bytesPerDomainShardedStore,
Transformer: transformer.TransformFunc(flattenIntoBytesPerDevice),
Writer: bytesPerDomainPerDeviceStore,
},
transformer.PipelineStage{
Name: "FlattenIntoBytesPerTimestamp",
Reader: bytesPerDomainShardedStore,
Transformer: transformer.TransformFunc(flattenIntoBytesPerTimestamp),
Writer: bytesPerDomainStore,
},
transformer.PipelineStage{
Name: "BytesPerDomainPostgresStore",
Reader: bytesPerDomainStore,
Writer: bytesPerDomainPostgresStore,
},
}, TraceKeyRangesPipeline(newTracesStore, traceKeyRangesStore, consolidatedTraceKeyRangesStore)...)
}