Merge pull request #4295 from vespa-engine/geirst/mmap-vs-directio-cl…

…oud-test

Prepare for mmap vs directio test in Vespa Cloud

tests/performance/mmap_vs_directio/app/services.xml

@@ -7,7 +7,7 @@
         <document-api/>
         <search/>
         <nodes deploy:environment="perf" deploy:region="aws-us-east-1c" count="1">
-            <resources vcpu="8.0" memory="16Gb" architecture="x86_64" storage-type="remote" disk="48Gb"/>
+            <resources vcpu="8.0" memory="16Gb" architecture="arm64" storage-type="remote" disk="48Gb"/>
         </nodes>
     </container>
 
@@ -18,7 +18,7 @@
         </documents>
         <nodes deploy:environment="perf" deploy:region="aws-us-east-1c" count="1">
             <!-- Memory available by proton: 12993349812 -->
-            <resources vcpu="8.0" memory="16Gb" architecture="x86_64" storage-type="local" disk="200Gb"/>
+            <resources vcpu="8.0" memory="16Gb" architecture="x86_64" storage-type="local" disk="474Gb"/>
         </nodes>
         <config name='vespa.config.search.core.proton'>
             <search>

tests/performance/mmap_vs_directio/create_updates.py

@@ -0,0 +1,38 @@
+# Copyright Vespa.ai. All rights reserved.
+
+import argparse
+import json
+import random
+import string
+import sys
+
+parser = argparse.ArgumentParser(description='Creates a JSON update for the payload field for each document id from standard in')
+parser.add_argument('--payload_bytes', type=int, default=600)
+parser.add_argument('--clear', action='store_true')
+args = parser.parse_args()
+
+payload_bytes = args.payload_bytes
+base64_length = int(payload_bytes * 8 / 6)
+num_payloads = 1000
+random.seed(1234)
+clear = args.clear
+
+payloads = [
+    ''.join(random.choices(string.ascii_letters + string.digits, k=base64_length))
+    for _ in range(num_payloads)
+]
+
+i = 0
+for line in sys.stdin:
+    docid = line.strip()
+    update = {
+        "update": docid,
+        "fields": {
+            "payload": {
+                "assign": None if clear else payloads[i % len(payloads)]
+            }
+        }
+    }
+    i += 1
+    print(json.dumps(update))
+