Share via


サーバーのパフォーマンス履歴

適用対象: Windows Server 2022、Windows Server 2019

記憶域スペース ダイレクトのパフォーマンス履歴」のサブトピックでは、サーバーから収集されるパフォーマンス履歴について詳しく説明します。 パフォーマンス履歴は、クラスター内のすべてのサーバーで利用できます。

注意

停止しているサーバーからパフォーマンス履歴を収集することはできません。 サーバーが復旧すると、自動的に収集が再開されます。

シリーズの名前と単位

以下のシリーズは、対象となるサーバーごとに収集されます。

系列 ユニット
clusternode.cpu.usage パーセント
clusternode.cpu.usage.guest パーセント
clusternode.cpu.usage.host パーセント
clusternode.memory.total バイト
clusternode.memory.available バイト
clusternode.memory.usage バイト
clusternode.memory.usage.guest バイト
clusternode.memory.usage.host バイト

さらに、サーバーに接続されているすべての対象ドライブのドライブ シリーズ physicaldisk.size.total が集計され、サーバーに接続されているすべての対象ネットワーク アダプターについて networkadapter.bytes.total などのネットワーク アダプター シリーズが集計されます。

解釈方法

系列 解釈方法
clusternode.cpu.usage アイドル状態ではないプロセッサ時間の割合。
clusternode.cpu.usage.guest ゲスト (仮想マシン) の要求に使用されたプロセッサ時間の割合。
clusternode.cpu.usage.host ホストの要求に使用されたプロセッサ時間の割合。
clusternode.memory.total サーバーの物理メモリの合計容量。
clusternode.memory.available サーバーで使用可能なメモリ。
clusternode.memory.usage サーバーで割り当てられている (使用できない) メモリ。
clusternode.memory.usage.guest ゲスト (仮想マシン) の要求に割り当てられたメモリ。
clusternode.memory.usage.host ホストの要求に割り当てられたメモリ。

取得元について

cpu.* シリーズは、Hyper-V が有効になっているかどうかに応じて、さまざまなパフォーマンス カウンターから収集されます。

Hyper-V が有効になっている場合:

系列 取得元のカウンター
clusternode.cpu.usage Hyper-V Hypervisor Logical Processor > _Total > % Total Run Time
clusternode.cpu.usage.guest Hyper-V Hypervisor Virtual Processor > _Total > % Total Run Time
clusternode.cpu.usage.host Hyper-V Hypervisor Root Virtual Processor > _Total > % Total Run Time

% Total Run Time カウンターを使用すると、パフォーマンス履歴属性のすべての使用量が保証されます。

Hyper-V が有効になっていない場合:

系列 取得元のカウンター
clusternode.cpu.usage Processor > _Total > % Processor Time
clusternode.cpu.usage.guest ゼロ
clusternode.cpu.usage.host 合計使用量と同じ

不完全同期にも関わらず、clusternode.cpu.usage は常に clusternode.cpu.usage.host + clusternode.cpu.usage.guest です。

同じように、clusternode.cpu.usage.guest は常にホスト サーバー上のすべての仮想マシンの vm.cpu.usage の合計になります。

memory.* シリーズは (近日公開予定) です。

注意

カウンターは、サンプリングされるのではなく、間隔全体で測定されます。 たとえば、サーバーが 9 秒間アイドル状態で、10 秒目に CPU 100% に急増した場合、その clusternode.cpu.usage はこの 10 秒間隔の平均で 10% として記録されます。 これにより、パフォーマンス履歴ですべてのアクティビティが把握され、履歴がノイズに対して堅牢になります。

PowerShell での使用法

Get-ClusterNode コマンドレットを使用します。

Get-ClusterNode <Name> | Get-ClusterPerf

その他の参照情報