merge: Support conversion from Azure traces to OpenDC traces (#36)

This pull request adds support for converting Azure traces to the OpenDC trace format.

* Support GZIP files in Azure trace 
* Fix timestamp retrieval for Azure trace
* Add column for CPU capacity in OpenDC format 
* Support conversion from Azure trace format 

3 files changed, 14 insertions, 9 deletions

diff --git a/opendc-trace/opendc-trace-azure/src/main/kotlin/org/opendc/trace/azure/AzureResourceStateTableReader.kt b/opendc-trace/opendc-trace-azure/src/main/kotlin/org/opendc/trace/azure/AzureResourceStateTableReader.kt
index da8181fe..94a91999 100644
--- a/opendc-trace/opendc-trace-azure/src/main/kotlin/org/opendc/trace/azure/AzureResourceStateTableReader.kt
+++ b/opendc-trace/opendc-trace-azure/src/main/kotlin/org/opendc/trace/azure/AzureResourceStateTableReader.kt
@@ -53,7 +53,7 @@ internal class AzureResourceStateTableReader(private val parser: CsvParser) : Ta
             when (parser.currentName) {
                 "timestamp" -> timestamp = Instant.ofEpochSecond(parser.longValue)
                 "vm id" -> id = parser.text
-                "CPU avg cpu" -> cpuUsagePct = parser.doubleValue
+                "CPU avg cpu" -> cpuUsagePct = (parser.doubleValue / 100.0) // Convert from % to [0, 1]
             }
         }
 
diff --git a/opendc-trace/opendc-trace-azure/src/main/kotlin/org/opendc/trace/azure/AzureResourceTableReader.kt b/opendc-trace/opendc-trace-azure/src/main/kotlin/org/opendc/trace/azure/AzureResourceTableReader.kt
index a6352613..6246dc35 100644
--- a/opendc-trace/opendc-trace-azure/src/main/kotlin/org/opendc/trace/azure/AzureResourceTableReader.kt
+++ b/opendc-trace/opendc-trace-azure/src/main/kotlin/org/opendc/trace/azure/AzureResourceTableReader.kt
@@ -52,8 +52,8 @@ internal class AzureResourceTableReader(private val parser: CsvParser) : TableRe
 
             when (parser.currentName) {
                 "vm id" -> id = parser.text
-                "vm created" -> startTime = Instant.ofEpochSecond(parser.longValue)
-                "vm deleted" -> stopTime = Instant.ofEpochSecond(parser.longValue)
+                "timestamp vm created" -> startTime = Instant.ofEpochSecond(parser.longValue)
+                "timestamp vm deleted" -> stopTime = Instant.ofEpochSecond(parser.longValue)
                 "vm virtual core count" -> cpuCores = parser.intValue
                 "vm memory" -> memCapacity = parser.doubleValue * 1e6 // GB to KB
             }
diff --git a/opendc-trace/opendc-trace-azure/src/main/kotlin/org/opendc/trace/azure/AzureTraceFormat.kt b/opendc-trace/opendc-trace-azure/src/main/kotlin/org/opendc/trace/azure/AzureTraceFormat.kt
index 253c7057..c9982877 100644
--- a/opendc-trace/opendc-trace-azure/src/main/kotlin/org/opendc/trace/azure/AzureTraceFormat.kt
+++ b/opendc-trace/opendc-trace-azure/src/main/kotlin/org/opendc/trace/azure/AzureTraceFormat.kt
@@ -31,8 +31,9 @@ import org.opendc.trace.util.CompositeTableReader
 import java.nio.file.Files
 import java.nio.file.Path
 import java.util.stream.Collectors
-import kotlin.io.path.extension
-import kotlin.io.path.nameWithoutExtension
+import java.util.zip.GZIPInputStream
+import kotlin.io.path.inputStream
+import kotlin.io.path.name
 
 /**
  * A format implementation for the Azure v1 format.
@@ -81,7 +82,10 @@ public class AzureTraceFormat : TraceFormat {
 
     override fun newReader(path: Path, table: String): TableReader {
         return when (table) {
-            TABLE_RESOURCES -> AzureResourceTableReader(factory.createParser(path.resolve("vmtable/vmtable.csv").toFile()))
+            TABLE_RESOURCES -> {
+                val stream = GZIPInputStream(path.resolve("vmtable/vmtable.csv.gz").inputStream())
+                AzureResourceTableReader(factory.createParser(stream))
+            }
             TABLE_RESOURCE_STATES -> newResourceStateReader(path)
             else -> throw IllegalArgumentException("Table $table not supported")
         }
@@ -96,8 +100,8 @@ public class AzureTraceFormat : TraceFormat {
      */
     private fun newResourceStateReader(path: Path): TableReader {
         val partitions = Files.walk(path.resolve("vm_cpu_readings"), 1)
-            .filter { !Files.isDirectory(it) && it.extension == "csv" }
-            .collect(Collectors.toMap({ it.nameWithoutExtension }, { it }))
+            .filter { !Files.isDirectory(it) && it.name.endsWith(".csv.gz") }
+            .collect(Collectors.toMap({ it.name.removeSuffix(".csv.gz") }, { it }))
             .toSortedMap()
         val it = partitions.iterator()
 
@@ -105,7 +109,8 @@ public class AzureTraceFormat : TraceFormat {
             override fun nextReader(): TableReader? {
                 return if (it.hasNext()) {
                     val (_, partPath) = it.next()
-                    return AzureResourceStateTableReader(factory.createParser(partPath.toFile()))
+                    val stream = GZIPInputStream(partPath.inputStream())
+                    return AzureResourceStateTableReader(factory.createParser(stream))
                 } else {
                     null
                 }