From: Laura Flores <lflores@redhat.com>
Date: Fri, 24 Dec 2021 01:18:13 +0000 (+0000)
Subject: os/bluestore: detect unnecessary zeros in _do_write_big()
X-Git-Tag: v17.1.0~159^2~1
X-Git-Url: http://git-server-git.apps.pok.os.sepia.ceph.com/?a=commitdiff_plain;h=914debdd4a5c3dd8f7b7762cd138e971a44cfdc3;p=ceph.git

os/bluestore: detect unnecessary zeros in _do_write_big()

Bluestore's `_do_write()` method handles writing data from bufferlists. Currently, it writes data from bufferlists without checking for unnecessary zeros. The lack zero detection may negatively impact performance.

In _do_write_big, we also check if a bufferlist is made up of zeros and avoid writing it if so.
Two new counters, `l_bluestore_write_big_skipped_blobs` and `l_bluestore_write_big_skipped_bytes`, have been introduced to help us count how many zero blocks and bytes from _do_write_big() have been skipped.

Signed-off-by: Laura Flores <lflores@redhat.com>
---

diff --git a/src/os/bluestore/BlueStore.cc b/src/os/bluestore/BlueStore.cc
index 8cf3bef1165c..dfb74df41420 100644
--- a/src/os/bluestore/BlueStore.cc
+++ b/src/os/bluestore/BlueStore.cc
@@ -5091,6 +5091,12 @@ void BlueStore::_init_logger()
 		    PerfCountersBuilder::PRIO_DEBUGONLY,
 		    unit_t(UNIT_BYTES));
 
+  b.add_u64_counter(l_bluestore_write_big_skipped_blobs,
+      "write_big_skipped_blobs",
+      "Large aligned writes into fresh blobs skipped due to zero detection (blobs)");
+  b.add_u64_counter(l_bluestore_write_big_skipped_bytes,
+      "write_big_skipped_bytes",
+      "Large aligned writes into fresh blobs skipped due to zero detection (bytes)");
   b.add_u64_counter(l_bluestore_write_small_skipped,
       "write_small_skipped",
       "Small writes into existing or sparse small blobs skipped due to zero detection");
@@ -14968,14 +14974,28 @@ void BlueStore::_do_write_big(
     }
     bufferlist t;
     blp.copy(l, t);
-    wctx->write(offset, b, l, b_off, t, b_off, l, false, new_blob);
-    dout(20) << __func__ << " schedule write big: 0x"
+
+    // Zero detection -- big block
+    if (!t.is_zero()) {
+      wctx->write(offset, b, l, b_off, t, b_off, l, false, new_blob);
+
+      dout(20) << __func__ << " schedule write big: 0x"
       << std::hex << offset << "~" << l << std::dec
       << (new_blob ? " new " : " reuse ")
       << *b << dendl;
+
+      logger->inc(l_bluestore_write_big_blobs);
+    } else { // if (!t.is_zero())
+      dout(20) << __func__ << " skip big zero block " << std::hex
+        << " (0x" << b_off << "~" << t.length() << ")"
+        << " (0x" << b_off << "~" << l << ")"
+        << std::dec << dendl;
+      logger->inc(l_bluestore_write_big_skipped_blobs);
+      logger->inc(l_bluestore_write_big_skipped_bytes, l);
+    }
+
     offset += l;
     length -= l;
-    logger->inc(l_bluestore_write_big_blobs);
   }
 }
 
diff --git a/src/os/bluestore/BlueStore.h b/src/os/bluestore/BlueStore.h
index d6ab76019656..f96b82b34c46 100644
--- a/src/os/bluestore/BlueStore.h
+++ b/src/os/bluestore/BlueStore.h
@@ -142,6 +142,8 @@ enum {
   l_bluestore_submitted_deferred_writes,
   l_bluestore_submitted_deferred_write_bytes,
 
+  l_bluestore_write_big_skipped_blobs,
+  l_bluestore_write_big_skipped_bytes,
   l_bluestore_write_small_skipped,
   l_bluestore_write_small_skipped_bytes,
   //****************************************