JP2010113383A

JP2010113383A - Storage system, method of determining performance thereof, and disk array control section

Info

Publication number: JP2010113383A
Application number: JP2008282747A
Authority: JP
Inventors: Manabu Miyazaki; 学宮崎
Original assignee: NEC Corp
Current assignee: NEC Corp
Priority date: 2008-11-04
Filing date: 2008-11-04
Publication date: 2010-05-20
Anticipated expiration: 2028-11-04
Also published as: JP5223602B2

Abstract

<P>PROBLEM TO BE SOLVED: To provide a storage system capable of determining the reference value of proper performance, a method of determining performance thereof, and a disk array control section. <P>SOLUTION: A disk array control unit 110 includes CPUs 111-11m and a statistical information storage means 130. The CPUs 111-11m includes performance determination means for determining the suitability of the configurations of logical disks 120-12n. The statistical information storage means 130 includes a reference response time determination means which inputs a load of an input/output command and determines an initial reference value determined from the statistical data obtained by measuring the process performance information and an estimated reference value determined by adding to the statistical data the process performance information when performing an input/output command process in a normal business. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

本発明は、ストレージシステム、その性能判定方法、ディスクアレイ制御部に関し、特に、蓄積した統計データを用いて、ハードウェア構成（以降、ＨＷ構成と称す）の適正性能の基準値を決定するストレージシステム、その性能判定方法、ディスクアレイ制御部に関する。 The present invention relates to a storage system, a performance determination method thereof, and a disk array control unit, and in particular, a storage system that determines a reference value of appropriate performance of a hardware configuration (hereinafter referred to as an HW configuration) using accumulated statistical data. The present invention relates to a performance determination method and a disk array controller.

任意のタイミング、所望の状況下でサーバシステムの性能を監視する技術が特許文献１に記載されている。 Patent Document 1 describes a technique for monitoring the performance of a server system at an arbitrary timing and a desired situation.

特許文献１記載の技術は、システム性能監視装置に設ける観測手段が、サーバの動作状態に関する観測情報を収集し、観測情報記憶手段に格納する。性能の測定に必要な観測情報に不足があれば、観測手段は、アクセス発生手段に観測情報を取得可能にする所定のアクセスを指示する。アクセス発生手段が所定のアクセスを発生させると、不足する観測情報に対応する観測可能状況が作り出され、観測情報が生成される。観測手段は、観測情報を収集し、観測情報記憶手段に格納する。観測情報記憶手段に観測情報が揃うと、性能評価手段は、必要な観測情報を読み出し、性能測定を行い性能基準を満たしているかどうかを評価する。 In the technique described in Patent Document 1, observation means provided in the system performance monitoring apparatus collects observation information regarding the operating state of the server and stores it in the observation information storage means. If there is a shortage of observation information necessary for performance measurement, the observation unit instructs the access generation unit to perform predetermined access that enables the observation information to be acquired. When the access generating means generates a predetermined access, an observable situation corresponding to the insufficient observation information is created, and observation information is generated. The observation means collects observation information and stores it in the observation information storage means. When the observation information is stored in the observation information storage means, the performance evaluation means reads out necessary observation information, performs performance measurement, and evaluates whether the performance standard is satisfied.

また、性能劣化したディスクの検出を行う技術が特許文献２に記載されている。 Patent Document 2 describes a technique for detecting a disk whose performance has deteriorated.

特許文献２記載の技術は、複数の物理ディスクを装備したディスクアレイ装置の各物理ディスクへのアクセスに対応する応答時間をそれぞれ計測する応答時間計測手段と、各物理ディスクの応答時間と、複数の物理ディスクのうち特定の物理ディスクの応答時間とに基づいて、特定の物理ディスクの性能劣化を判断する性能劣化判断手段とを設ける。この構成により、上記技術は、ディスクアレイ装置を構成する各物理ディスクの性能劣化を精度よく検出し、障害の発生を未然に防ぐものである。 The technique described in Patent Document 2 includes a response time measuring unit that measures response times corresponding to accesses to each physical disk of a disk array device equipped with a plurality of physical disks, a response time of each physical disk, and a plurality of response times. Performance degradation determining means for determining performance degradation of a specific physical disk based on the response time of the specific physical disk among the physical disks is provided. With this configuration, the above technique detects the performance deterioration of each physical disk constituting the disk array apparatus with high accuracy and prevents the occurrence of a failure in advance.

特開２００７−０２６３０３号公報JP 2007-026303 A 特開２００６−０９２０７０号公報JP 2006-092070 A

性能基準値の設定は、設定者の経験によって決められる場合、主観的な決定方法であるため、設定者のスキルに依存されることになる。この場合、性能基準値は、システムの実情とはかけ離れた値に設定される可能性も排除できない。 Since the setting of the performance reference value is a subjective determination method when it is determined by the experience of the setter, it depends on the skill of the setter. In this case, the possibility that the performance reference value is set to a value far from the actual situation of the system cannot be excluded.

上記特許文献１は、性能測定結果と予め決められた性能基準とを比較している。従って、性能基準値は、精度の低い設定値となる可能性がある。性能は、負荷の種類、ＨＷコンポーネントの種類によって異なるため、適切な性能基準値の設定は、難しく容易ではないという課題がある。 Patent Document 1 compares a performance measurement result with a predetermined performance standard. Therefore, the performance reference value may be a setting value with low accuracy. Since the performance varies depending on the type of load and the type of HW component, there is a problem that setting an appropriate performance reference value is difficult and not easy.

また、上記特許文献２は、各物理ディスクの応答時間と、複数の物理ディスクのうち特定の物理ディスクの応答時間とに基づいて、特定の物理ディスクの性能劣化を判断している。即ち、性能基準値に代え平均応答時間と、特定の物理ディスクの応答時間とを比較し、性能劣化を判断している。従って、ディスクアレイ装置を構成する物理ディスクの台数が少ないとき、１台の物理ディスクの性能劣化に伴い平均応答時間の変化が大きくなり、特定の物理ディスクの性能劣化の判断が正確に行えないという課題がある。 Further, Patent Document 2 determines the performance deterioration of a specific physical disk based on the response time of each physical disk and the response time of a specific physical disk among a plurality of physical disks. That is, instead of the performance reference value, the average response time is compared with the response time of a specific physical disk to determine performance degradation. Therefore, when the number of physical disks that make up the disk array device is small, the change in average response time increases with the performance deterioration of one physical disk, and it is not possible to accurately determine the performance deterioration of a specific physical disk. There are challenges.

本発明の目的は、上述の課題を解決すべく適正な性能の基準値を決定することのできるストレージシステム、その性能判定方法、ディスクアレイ制御部を提供することにある。 An object of the present invention is to provide a storage system capable of determining an appropriate performance reference value to solve the above-described problems, a performance determination method thereof, and a disk array control unit.

本発明のストレージシステムは、入出力コマンドを発行するホストコンピュータと入出力コマンド処理を実行するストレージ装置とを接続したストレージシステムであって、ストレージ装置は、複数の論理ディスクと、論理ディスクを制御するディスクアレイ制御部とを有し、ディスクアレイ制御部は、論理ディスクの処理性能情報を収集する性能情報採取手段と、論理ディスクの構成の適否を判定する性能判定手段と、処理性能情報を蓄積する統計情報蓄積手段とを有し、統計情報蓄積手段は、入出力コマンドの負荷を投入し処理性能情報を実測した統計データから求める初期基準値と通常業務上で入出力コマンド処理を実行したときの処理性能情報を統計データに追加することで求める予測基準値とを決定する基準応答時間決定手段を有することを特徴とする。 The storage system of the present invention is a storage system in which a host computer that issues input / output commands and a storage device that executes input / output command processing are connected, and the storage device controls a plurality of logical disks and logical disks. A disk array controller, and the disk array controller stores performance information collecting means for collecting logical disk processing performance information, performance determining means for determining whether the logical disk configuration is appropriate, and storing processing performance information. Statistical information storage means, the statistical information storage means, when the input / output command load is input and the processing performance information is obtained from the measured statistical data, the initial reference value and when the input / output command processing is executed on normal business There is a reference response time determination means that determines the predicted reference value obtained by adding processing performance information to statistical data. And wherein the Rukoto.

本発明のストレージシステムの性能判定方法は、入出力コマンドを発行するホストコンピュータと入出力コマンド処理を実行するストレージ装置とを接続したストレージシステムの性能判定方法であって、ストレージ装置は、複数の論理ディスクと、論理ディスクを制御するディスクアレイ制御部とを有し、ディスクアレイ制御部が、論理ディスクの処理性能情報を収集するステップと、論理ディスクの構成の適否を性能判定手段により判定するステップと、処理性能情報を統計情報蓄積手段へ蓄積するステップとを有し、統計情報蓄積手段が、入出力コマンドの負荷を投入し処理性能情報を実測した統計データから初期基準値を求めるステップと、通常業務上で入出力コマンド処理を実行したときの処理性能情報を統計データに追加することで予測基準値を求めるステップとを有することを特徴とする。 The storage system performance determination method of the present invention is a storage system performance determination method in which a host computer that issues input / output commands and a storage apparatus that executes input / output command processing are connected. A disk and a disk array control unit that controls the logical disk, the disk array control unit collecting logical disk processing performance information, and determining whether or not the logical disk configuration is appropriate by a performance determination unit A step of accumulating processing performance information in the statistical information accumulating means, wherein the statistical information accumulating means obtains an initial reference value from statistical data obtained by measuring the processing performance information by applying a load of an input / output command; Add processing performance information to the statistical data when I / O command processing is executed on the job In and having a step of obtaining a predicted reference value.

本発明のディスクアレイ制御部は、複数の論理ディスクを制御するディスクアレイ制御部であって、複数のＣＰＵ（ＣｅｎｔｒａｌＰｒｏｃｅｓｓｉｎｇＵｎｉｔ）と、論理ディスクの処理性能情報を蓄積する統計情報蓄積手段とを有し、ＣＰＵは、論理ディスクへの入出力コマンド処理を行うＩ／Ｏ（Ｉｎｐｕｔ／Ｏｕｔｐｕｔ）処理手段と、論理ディスクの処理性能情報を収集する性能情報採取手段と、論理ディスクの構成の適否を判定する性能判定手段と、ＨＷ構成適否判定手段により不適と判断したＨＷ構成を変更するＨＷ構成変更手段とを有し、統計情報蓄積手段は、入出力コマンドの負荷を投入し処理性能情報を実測した統計データから求める初期基準値と通常業務上で入出力コマンド処理を実行したときの処理性能情報を統計データに追加することで求める予測基準値とを決定する基準応答時間決定手段と、初期基準値を求めるための初期表を作成する初期化手段と、初期表から入出力コマンド数の性能限界を求める限界Ｉ／Ｏ処理数判別手段と、初期表に統計データを追加する統計データ追加手段とを有することを特徴とする。 The disk array control unit of the present invention is a disk array control unit that controls a plurality of logical disks, and has a plurality of CPUs (Central Processing Units) and statistical information storage means for storing processing performance information of the logical disks. The CPU determines whether an I / O (Input / Output) processing unit that performs input / output command processing to the logical disk, a performance information collection unit that collects processing performance information of the logical disk, and whether or not the logical disk configuration is appropriate. And a HW configuration change unit that changes the HW configuration determined to be inappropriate by the HW configuration suitability determination unit, and the statistical information storage unit actually measured the processing performance information by applying the input / output command load. The initial reference value obtained from statistical data and the processing performance information when I / O command processing is executed in normal business are integrated. The reference response time determination means for determining the predicted reference value to be obtained by adding to the total data, the initialization means for creating the initial table for obtaining the initial reference value, and the performance limit of the number of input / output commands from the initial table It has a limit I / O processing number determination means to be obtained, and statistical data addition means for adding statistical data to the initial table.

本発明のストレージシステム、ストレージシステム性能判定方法、ディスクアレイ制御部は、統計情報蓄積手段から予測基準値を決定することにより、実測値に基づいた客観的で適切な性能基準値を設定できるという効果がある。 The storage system, the storage system performance determination method, and the disk array control unit of the present invention can set an objective and appropriate performance reference value based on an actual measurement value by determining a prediction reference value from the statistical information storage unit. There is.

次に、本発明の第１の実施の形態について図面を参照して詳細に説明する。 Next, a first embodiment of the present invention will be described in detail with reference to the drawings.

図１は、本発明のストレージシステム１００を示す概略構成ブロック図、図２は、図１中の統計情報蓄積手段１３０の詳細を示す概略構成ブロック図、図３は、図１中のＣＰＵ１１１〜１１ｍ（但し、ｍは２以上の自然数）の詳細を示す概略構成ブロック図である。 1 is a schematic configuration block diagram showing a storage system 100 of the present invention, FIG. 2 is a schematic configuration block diagram showing details of the statistical information storage means 130 in FIG. 1, and FIG. 3 is a CPU 111-11m in FIG. It is a schematic block diagram showing details of (where m is a natural number of 2 or more).

図１を参照すると、ストレージシステム１００は、入出力コマンドを発行するホストコンピュータ２０１と入出力コマンド処理を実行するストレージ装置１０１とを接続して構成する。 Referring to FIG. 1, a storage system 100 is configured by connecting a host computer 201 that issues input / output commands and a storage apparatus 101 that executes input / output command processing.

ストレージ装置１０１は、複数の論理ディスク１２０〜１２ｎ（但し、ｎは自然数）と、論理ディスク１２０〜１２ｎを制御するディスクアレイ制御部１１０とで構成する。 The storage apparatus 101 includes a plurality of logical disks 120 to 12n (where n is a natural number) and a disk array control unit 110 that controls the logical disks 120 to 12n.

ホストコンピュータ２０１とストレージ装置１０１とは、ホストバスアダプタ２１１とディスクアレイ制御部１１０とで接続する。 The host computer 201 and the storage apparatus 101 are connected by the host bus adapter 211 and the disk array control unit 110.

ディスクアレイ制御部１１０は、複数のＣＰＵ１１１〜１１ｍと、論理ディスク１２０〜１２ｎの処理性能情報を蓄積する統計情報蓄積手段１３０とを備える。 The disk array control unit 110 includes a plurality of CPUs 111 to 11m and statistical information storage means 130 that stores processing performance information of the logical disks 120 to 12n.

図３を参照すると、ＣＰＵ１１１〜１１ｍは、Ｉ／Ｏ処理手段１５０と、性能情報採取手段１４０と、性能判定手段１６０と、ＨＷ構成変更手段１８０とを備える。 Referring to FIG. 3, the CPUs 111 to 11 m include an I / O processing unit 150, a performance information collection unit 140, a performance determination unit 160, and a HW configuration change unit 180.

Ｉ／Ｏ処理手段１５０は、論理ディスク１２０〜１２ｎへの入出力コマンド処理を行う。 The I / O processing means 150 performs input / output command processing for the logical disks 120 to 12n.

性能情報採取手段１４０は、論理ディスク１２０〜１２ｎの処理性能情報を収集する。処理性能情報は、、単位時間当たりの入出力コマンド数（Ｉ／Ｏ数／ｓｅｃ１４１）と、入出力コマンド数に対する応答時間１４２とを備える。 The performance information collecting unit 140 collects processing performance information of the logical disks 120 to 12n. The processing performance information includes the number of input / output commands per unit time (I / O number / sec 141) and a response time 142 for the number of input / output commands.

性能判定手段１６０は、論理ディスク１２０〜１２ｎの構成の適否を判定するＨＷ構成適否判定手段１６１を備える。ＨＷ構成適否判定手段１６１は、論理ディスク構成１６２の適否判定と、ＣＰＵ構成１６３の適否判定とを行う。 The performance determination unit 160 includes a HW configuration suitability determination unit 161 that determines whether the configuration of the logical disks 120 to 12n is appropriate. The HW configuration suitability determination unit 161 performs suitability determination of the logical disk configuration 162 and suitability determination of the CPU configuration 163.

ＨＷ構成変更手段１８０は、ＨＷ構成適否判定手段１６１により不適と判断したＨＷ構成を変更する。 The HW configuration change unit 180 changes the HW configuration determined as inappropriate by the HW configuration suitability determination unit 161.

図２を参照すると、統計情報蓄積手段１３０は、初期化手段１３１と、統計データ追加手段１３２と、限界Ｉ／Ｏ処理数判別手段１３３と、基準応答時間決定手段１３４とを備える。 Referring to FIG. 2, the statistical information accumulation unit 130 includes an initialization unit 131, a statistical data addition unit 132, a limit I / O processing number determination unit 133, and a reference response time determination unit 134.

初期化手段１３１は、初期基準値１３５を求めるための初期表１７４を作成する。統計データ追加手段１３２は、初期表１７４に統計データを追加する。限界Ｉ／Ｏ処理数判別手段１３３は、初期表１７４から入出力コマンド数の性能限界１７１を求める。 The initialization unit 131 creates an initial table 174 for obtaining the initial reference value 135. The statistical data adding unit 132 adds statistical data to the initial table 174. The limit I / O processing number determination unit 133 obtains the performance limit 171 of the number of input / output commands from the initial table 174.

基準応答時間決定手段１３４は、初期基準値１３５と予測基準値１３６とを決定する。初期基準値１３５は、入出力コマンドの負荷を投入し、処理性能情報を実測した統計データから求める。予測基準値１３６は、通常業務上で入出力コマンド処理を実行したときの処理性能情報を統計データに追加することで求める。 The reference response time determining unit 134 determines the initial reference value 135 and the predicted reference value 136. The initial reference value 135 is obtained from statistical data obtained by actually measuring the processing performance information by inputting the input / output command load. The prediction reference value 136 is obtained by adding processing performance information when the input / output command processing is executed in the normal business to the statistical data.

次に、上述のように構成したストレージシステム１００の動作について説明する。 Next, the operation of the storage system 100 configured as described above will be described.

図１〜３を参照すると、ホストコンピュータ２０１の発行した入出力コマンドには、コマンド処理対象となる論理ディスク１２０〜１２ｎが指定されている。各々の論理ディスク１２０〜１２ｎには、コマンド処理を担当するＣＰＵ１１１〜１１ｍが対応付けされている。ＣＰＵ１１１〜１１ｍは、担当する論理ディスク１２０〜１２ｎへ要求された入出力コマンド処理を実行する。 Referring to FIGS. 1 to 3, the input / output commands issued by the host computer 201 specify logical disks 120 to 12n that are subject to command processing. Each of the logical disks 120 to 12n is associated with CPUs 111 to 11m in charge of command processing. The CPUs 111 to 11m execute input / output command processing requested to the logical disks 120 to 12n in charge.

ＣＰＵ１１１〜１１ｍは、処理実行時の単位時間当たりの入出力コマンド数と、入出力コマンド数に対する応答時間１４２との処理性能情報を収集し、収集した情報を統計情報蓄積手段１３０へ格納する。処理性能情報は、ホストコンピュータ２０１の発行した入出力コマンドが処理されるたびに統計情報蓄積手段１３０へ追加して格納される。 The CPUs 111 to 11 m collect processing performance information about the number of input / output commands per unit time at the time of processing execution and the response time 142 with respect to the number of input / output commands, and store the collected information in the statistical information storage unit 130. The processing performance information is additionally stored in the statistical information storage unit 130 each time an input / output command issued by the host computer 201 is processed.

処理性能情報を統計情報蓄積手段１３０へ格納後、ＣＰＵ１１１〜１１ｍの性能判定手段１６０は、現在のＨＷ構成が処理した入出力コマンド数に対する応答時間１４２を元に適切かどうかを判断する。ここに、性能判定手段１６０は、予測基準値１３６の応答時間１４２と、現在のＨＷ構成が処理した応答時間１４２とを比較して判定を行う。 After the processing performance information is stored in the statistical information storage unit 130, the performance determination unit 160 of the CPUs 111 to 11m determines whether or not the current HW configuration is appropriate based on the response time 142 for the number of input / output commands processed. Here, the performance determination means 160 makes a determination by comparing the response time 142 of the prediction reference value 136 with the response time 142 processed by the current HW configuration.

例えば、論理ディスク１２０の性能が足りず適切でないと性能判定手段１６０が判断したとき、ＨＷ構成変更手段１８０は、物理ディスク１２０−１を論理ディスク１２０に追加する。なお、ＣＰＵ１１１〜１１ｍの性能が足りないと性能判定手段１６０が判断したとき、ＨＷ構成変更手段１８０は、ＣＰＵ１１１〜１１ｍに割り当てられている論理ディスク１２０〜１２ｎの数を減らす変更を行う。 For example, when the performance determination unit 160 determines that the performance of the logical disk 120 is insufficient and inappropriate, the HW configuration change unit 180 adds the physical disk 120-1 to the logical disk 120. When the performance determination unit 160 determines that the performance of the CPUs 111 to 11m is insufficient, the HW configuration change unit 180 performs a change to reduce the number of logical disks 120 to 12n allocated to the CPUs 111 to 11m.

次に、ＨＷ構成が適正性能か否かの判断を行うための予測基準値１３６の求め方について、図面を参照して説明する。 Next, how to obtain the prediction reference value 136 for determining whether or not the HW configuration has appropriate performance will be described with reference to the drawings.

図４は、入出力コマンドの負荷を投入し、処理性能情報を実測した統計データの初期表１７４を示す図、図５は、通常業務上で入出力コマンド処理を実行したときの処理性能情報を初期表１７４に追加した実行処理性能表１７５を示す図である。 FIG. 4 is a diagram showing an initial table 174 of statistical data in which input / output command load is input and processing performance information is actually measured. FIG. 5 shows processing performance information when the input / output command processing is executed in a normal operation. It is a figure which shows the execution processing performance table 175 added to the initial table 174. FIG.

応答時間１４２は、入出力コマンド数（Ｉ／Ｏ数／ｓｅｃ１４１）を多くし、負荷を増加させていくと、急激に長くなる（論理ディスク１２０〜１２ｎの性能が悪化することに起因）特性がある。 The response time 142 increases rapidly when the number of input / output commands (I / O number / sec 141) is increased and the load is increased (because the performance of the logical disks 120 to 12n deteriorates). is there.

図４を参照すると、負荷（入出力コマンド数）が論理ディスク１２０〜１２ｎの性能限界１７１に近づくと、応答時間１４２は急上昇する。 Referring to FIG. 4, when the load (number of input / output commands) approaches the performance limit 171 of the logical disks 120 to 12n, the response time 142 increases rapidly.

性能限界１７１は、ストレージ装置１０１のＨＷ構成（論理ディスク１２０〜１２ｎの数、論理ディスク１２０〜１２ｎを構成している物理ディスク１２０−１の数、物理ディスク１２０−１自体の性能）によって変化する。従って、入出力コマンド数、応答時間１４２の性能限界１７１を予め知ることは難しい。 The performance limit 171 changes depending on the HW configuration of the storage apparatus 101 (number of logical disks 120 to 12n, number of physical disks 120-1 constituting the logical disks 120 to 12n, performance of the physical disk 120-1 itself). . Therefore, it is difficult to know the number of input / output commands and the performance limit 171 of the response time 142 in advance.

そこで、先ず、ＨＷ構成の変更後に典型的な負荷を投入して、処理性能情報を実測し、図４に示すような初期表１７４を作成する。なお、投入する試験用負荷は、例えば、Ｉ／Ｏサイズ８ＫＢ、Ｒｅａｄ−Ｗｒｉｔｅの比率２：１、ランダムアクセスなどとする。 Therefore, first, after changing the HW configuration, a typical load is input, and the processing performance information is actually measured, and an initial table 174 as shown in FIG. 4 is created. Note that the test load to be input is, for example, an I / O size of 8 KB, a read-write ratio of 2: 1, random access, and the like.

予測基準値１３６の求め方に先んじて、図４の初期表１７４を参照して、初期基準値１３５（Ｔ０）の求め方について説明する。 Prior to obtaining the prediction reference value 136, the method for obtaining the initial reference value 135 (T0) will be described with reference to the initial table 174 of FIG.

初期表１７４を参照すると、現在のＨＷ構成の性能限界１７１にマージンを持たせたプレ性能限界１７２は、例えば、限界Ｉ／Ｏ処理数１７３の８０％に設定すると、致命的な性能劣化（応答時間１４２の急上昇）の発生を回避することができる。なお、プレ性能限界１７２は、応答時間１４２に変化があり、且つ、致命的な性能劣化に至らない入出力コマンド数（Ｎｐ）に設定する。 Referring to the initial table 174, if the pre-performance limit 172 having a margin for the performance limit 171 of the current HW configuration is set to 80% of the limit I / O processing count 173, for example, fatal performance degradation (response Occurrence of the rapid increase in time 142) can be avoided. The pre-performance limit 172 is set to the number of input / output commands (Np) that has a change in the response time 142 and does not lead to fatal performance degradation.

初期基準値１３５（Ｔ０）は、入出力コマンド数に対する応答時間１４２の変化開始時の入出力コマンド数（Ｎｓ）、応答時間１４２（Ｔｓ）と変化開始後の入出力コマンド数（Ｎａ）、応答時間１４２（Ｔａ）とを結んだ直線（図４中、黒丸印を結んだ直線）と、性能限界１７１にマージンを持たせたプレ性能限界１７２の入出力コマンド数（Ｎｐ）との交点の応答時間１４２（Ｔ０）とすることで求める。なお、変化開始時の入出力コマンド数（Ｎｓ）と変化開始後の入出力コマンド数（Ｎａ）とは、プレ性能限界１７２の入出力コマンド数（Ｎｐ）に最も近い両側の入出力コマンド数（Ｎｓ、Ｎａ）とする。 The initial reference value 135 (T0) includes the number of input / output commands (Ns) at the start of the change in the response time 142 with respect to the number of input / output commands, the response time 142 (Ts), the number of input / output commands after the change starts (Na), and the response Response of the intersection of the straight line connecting the time 142 (Ta) (the straight line connecting the black circles in FIG. 4) and the number of input / output commands (Np) of the pre-performance limit 172 with a margin for the performance limit 171 It is obtained by setting time 142 (T0). The number of input / output commands (Ns) at the start of change and the number of input / output commands (Na) after the start of change are the number of input / output commands on both sides closest to the number of input / output commands (Np) at the pre-performance limit 172 ( Ns, Na).

次に、予測基準値１３６の求め方について、図５を参照して説明する。 Next, how to obtain the prediction reference value 136 will be described with reference to FIG.

図５に示す実行処理性能表１７５を参照すると、予測基準値１３６は、統計データに追加した処理性能情報の内プレ性能限界１７２に最も近い入出力コマンド数（Ｎｎ）での応答時間１４２（Ｔｎ）と変化開始後の応答時間１４２（Ｔａ）とを結んだ直線（図５中、黒丸印を結んだ直線）と、プレ性能限界１７２の入出力コマンド数（Ｎｐ）との交点の応答時間１４２（Ｔｐ）とすることで求める。 Referring to the execution processing performance table 175 shown in FIG. 5, the prediction reference value 136 is a response time 142 (Tn) at the number of input / output commands (Nn) closest to the pre-performance limit 172 in the processing performance information added to the statistical data. ) And the response time 142 (Ta) after the start of change (the straight line connecting the black circles in FIG. 5) and the response time 142 at the intersection of the number of input / output commands (Np) of the pre-performance limit 172 It is obtained by setting (Tp).

次に、ＨＷ構成変更処理の動作について図面を参照して説明する。 Next, the operation of the HW configuration change process will be described with reference to the drawings.

図６は、ＨＷ構成変更処理の動作を説明するためのフローチャートである。 FIG. 6 is a flowchart for explaining the operation of the HW configuration change process.

図６を参照すると、ディスクアレイ制御部１１０は、論理ディスク１２０〜１２ｎを構築し（Ｓ１１）、負荷を投入して、論理ディスク１２０〜１２ｎの処理性能情報を実測し、図４に示すような初期表１７４を作成する（Ｓ１２）。 Referring to FIG. 6, the disk array control unit 110 constructs the logical disks 120 to 12n (S11), inputs a load, and actually measures the processing performance information of the logical disks 120 to 12n, as shown in FIG. An initial table 174 is created (S12).

このとき、ディスクアレイ制御部１１０は、初期表１７４を元にプレ性能限界１７２の入出力コマンド数（Ｎｐ）、初期基準値１３５の応答時間１４２（Ｔ０）、変化開始時の入出力コマンド数（Ｎｓ）、応答時間１４２（Ｔｓ）、変化開始後の入出力コマンド数（Ｎａ）、応答時間１４２（Ｔａ）を求めておく。 At this time, the disk array control unit 110, based on the initial table 174, the number of input / output commands of the pre-performance limit 172 (Np), the response time 142 (T0) of the initial reference value 135, the number of input / output commands at the start of change ( Ns), response time 142 (Ts), the number of input / output commands (Na) after the start of change, and response time 142 (Ta).

その後、ディスクアレイ制御部１１０は、通常業務の入出力コマンド処理を実行し（Ｓ１３）、処理を実行するたびに処理性能を測定し（Ｓ１４）、統計情報蓄積手段１３０へ追加して格納する（Ｓ１５）。 Thereafter, the disk array control unit 110 executes input / output command processing for normal business (S13), measures the processing performance each time the processing is executed (S14), and additionally stores it in the statistical information storage means 130 (S14). S15).

ディスクアレイ制御部１１０の性能判定手段１６０は、予測基準値１３６の応答時間１４２（Ｔｎ）と、現在のＨＷ構成が処理した応答時間１４２とを比較して、ＨＷ構成が負荷に対して適正性能か否かの判定を行う（Ｓ１６）。 The performance determination unit 160 of the disk array control unit 110 compares the response time 142 (Tn) of the prediction reference value 136 with the response time 142 processed by the current HW configuration, and the HW configuration has an appropriate performance with respect to the load. Is determined (S16).

ＨＷ構成が適正性能であるか否かを判定する予測基準値１３６の応答時間１４２（Ｔｎ）は、図５のように初期表１７４に処理性能情報が追加されると、初期基準値１３５の応答時間１４２（Ｔ０）から変更され、より精度が向上した正確な値に更新されていく。 The response time 142 (Tn) of the prediction reference value 136 for determining whether or not the HW configuration has proper performance is the response of the initial reference value 135 when the processing performance information is added to the initial table 174 as shown in FIG. It is changed from time 142 (T0) and updated to an accurate value with improved accuracy.

ＨＷ構成が負荷に対して適正であれば、上記（Ｓ1３）に戻り、適正でないとき、ディスクアレイ制御部１１０は、ＨＷ構成を変更する（Ｓ１７）。 If the HW configuration is appropriate for the load, the process returns to the above (S13). If it is not appropriate, the disk array control unit 110 changes the HW configuration (S17).

以上説明したように、ディスクアレイ制御部１１０は、統計情報蓄積手段１３０から予測基準値１３６を決定することにより、実測値に基づいた客観的で適切な性能の予測基準値１３６を設定できるという効果がある。 As described above, the disk array control unit 110 determines the prediction reference value 136 from the statistical information storage unit 130, so that the objective and appropriate performance prediction reference value 136 based on the actual measurement value can be set. There is.

また、ストレージシステム１００は、負荷に対して最適なＨＷ構成を決定することができるという効果がある。 In addition, the storage system 100 has an effect of being able to determine an optimal HW configuration for the load.

なお、上記実施の形態においては、統計情報として、単位時間当たりの入出力コマンド数（Ｉ／Ｏ数／ｓｅｃ１４１）と、入出力コマンド数に対する応答時間１４２とを用いているが、、入出力コマンド数に代えデータ転送量としてもよいことは云うまでもない。 In the above embodiment, the number of input / output commands per unit time (I / O number / sec 141) and the response time 142 for the number of input / output commands are used as statistical information. It goes without saying that the data transfer amount may be used instead of the number.

本発明のストレージシステムを示す概略構成ブロック図である。1 is a schematic configuration block diagram showing a storage system of the present invention. 図１中の統計情報蓄積手段の詳細を示す概略構成ブロック図である。FIG. 2 is a block diagram of a schematic configuration showing details of statistical information storage means in FIG. 1. 図１中のＣＰＵの詳細を示す概略構成ブロック図である。FIG. 2 is a schematic configuration block diagram showing details of a CPU in FIG. 1. 処理性能情報を実測した統計データの初期表を示す図である。It is a figure which shows the initial table of the statistical data which measured process performance information. 処理性能情報を初期表に追加した実行処理性能表を示す図である。It is a figure which shows the execution processing performance table which added processing performance information to the initial table. ＨＷ構成変更処理の動作を説明するためのフローチャートである。It is a flowchart for demonstrating operation | movement of a HW structure change process.

Explanation of symbols

１００ストレージシステム
１０１ストレージ装置
１１０ディスクアレイ制御部
１１１〜１１ｍＣＰＵ
１２０〜１２ｎ論理ディスク
１２０−１物理ディスク
１３０統計情報蓄積手段
１３１初期化手段
１３２統計データ追加手段
１３３限界Ｉ／Ｏ処理数判別手段
１３４基準応答時間決定手段
１３５初期基準値
１３６予測基準値
１４０性能情報採取手段
１４１Ｉ／Ｏ数／ｓｅｃ
１４２応答時間
１６０性能判定手段
１６１ＨＷ構成適否判定手段
１６２論理ディスク構成
１６３ＣＰＵ構成
１７１性能限界
１７２プレ性能限界
１７３限界Ｉ／Ｏ処理数
１７４初期表
１７５実行処理性能表
１８０ＨＷ構成変更手段
２０１ホストコンピュータ
２１１ホストバスアダプタ DESCRIPTION OF SYMBOLS 100 Storage system 101 Storage apparatus 110 Disk array control part 111-11m CPU
120 to 12n Logical disk 120-1 Physical disk 130 Statistical information storage means 131 Initialization means 132 Statistical data addition means 133 Limit I / O processing number determination means 134 Reference response time determination means 135 Initial reference value 136 Predictive reference value 140 Performance information Sampling means 141 I / O count / sec
142 Response Time 160 Performance Judgment Unit 161 HW Configuration Suitability Judgment Unit 162 Logical Disk Configuration 163 CPU Configuration 171 Performance Limit 172 Pre-Performance Limit 173 Limit I / O Processing Number 174 Initial Table 175 Execution Processing Performance Table 180 HW Configuration Change Unit 201 Host Computer 211 Host bus adapter

Claims

A storage system in which a host computer that issues an input / output command and a storage device that executes the input / output command processing are connected, the storage device including a plurality of logical disks and a disk array control unit that controls the logical disks And the disk array control unit accumulates the processing performance information, a performance information collecting unit that collects processing performance information of the logical disk, a performance determination unit that determines suitability of the configuration of the logical disk, and Statistical information storage means, and the statistical information storage means performs the input / output command processing on normal operations and an initial reference value obtained from statistical data obtained by actually measuring the processing performance information by applying a load of the input / output commands. A base for determining a prediction reference value to be obtained by adding the processing performance information at the time of execution to the statistical data Storage system characterized in that it has a response time determining means.

The storage system according to claim 1, wherein the processing performance information includes the number of input / output commands per unit time and a response time with respect to the number of input / output commands.

The storage system according to claim 1, wherein the performance determination unit performs the determination by comparing the prediction reference value with the processing performance information.

The statistical information storage means includes an initialization means for creating an initial table for obtaining the initial reference value, a limit I / O processing number determination means for obtaining a performance limit of the number of input / output commands from the initial table, The storage system according to any one of claims 1 to 3, further comprising statistical data adding means for adding statistical data to the initial table.

The initial reference value is a straight line connecting the response time at the start of change of the response time with respect to the number of input / output commands and after the start of change, and the input / output of the pre-performance limit with a margin for the performance limit. 5. The storage system according to claim 4, wherein the response time at the intersection with the number of commands is used.

The prediction reference value is a straight line connecting the response time at the number of input / output commands closest to the pre-performance limit in the processing performance information added to the statistical data and the response time after the start of change. 6. The storage system according to claim 5, wherein the response time is an intersection of the pre-performance limit and the number of input / output commands.

4. The storage according to claim 1, wherein the performance determination unit includes an HW configuration suitability determination unit configured to determine whether or not the logical disk configuration is appropriate and whether the CPU configuration is appropriate. system.

The disk array control unit includes a plurality of CPUs, and the CPU performs I / O processing means for performing the input / output command processing on the logical disk, the performance information collecting means, the performance determining means, 4. The storage system according to claim 1, further comprising an HW configuration change unit that changes an HW configuration that is determined to be inappropriate by the HW configuration suitability determination unit.

A performance determination method for a storage system in which a host computer that issues an input / output command and a storage apparatus that executes the input / output command process are connected, the storage apparatus controlling a plurality of logical disks and the logical disks A disk array control unit, wherein the disk array control unit collects processing performance information of the logical disk, determines whether or not the configuration of the logical disk is appropriate by a performance determination unit, and the processing performance information A statistical information accumulating means, wherein the statistical information accumulating means obtains an initial reference value from statistical data obtained by actually applying the input / output command load and measuring the processing performance information; The processing performance information when the input / output command processing is executed above is added to the statistical data. Performance determination method for a storage system, characterized by a step of obtaining a predicted reference value by.

10. The storage system performance determination method according to claim 9, wherein the processing performance information includes the number of input / output commands per unit time and a response time corresponding to the number of input / output commands.

The storage system performance determination method according to claim 9 or 10, wherein the performance determination unit performs the determination by comparing the prediction reference value and the processing performance information.

The statistical information storage means creating an initial table for obtaining the initial reference value; obtaining a performance limit of the number of input / output commands from the initial table; and adding statistical data to the initial table 12. The storage system performance determination method according to claim 9, further comprising:

The initial reference value is a straight line connecting the response time at the start of change of the response time with respect to the number of input / output commands and after the start of change, and the input / output of the pre-performance limit with a margin for the performance limit. 13. The storage system performance determination method according to claim 12, wherein the response time at the intersection with the number of commands is used.

The prediction reference value is a straight line connecting the response time at the number of input / output commands closest to the pre-performance limit in the processing performance information added to the statistical data and the response time after the start of change. 14. The storage system performance determination method according to claim 13, wherein the response time at the intersection of the pre-performance limit and the number of input / output commands is used.

A disk array control unit for controlling a plurality of logical disks, comprising: a plurality of CPUs; and statistical information storage means for storing processing performance information of the logical disks, wherein the CPUs input the logical disks. I / O processing means for performing output command processing, performance information collecting means for collecting the processing performance information of the logical disk, performance determining means for determining suitability of the logical disk configuration, and HW configuration suitability determining means An HW configuration changing unit that changes an HW configuration determined to be inappropriate, and the statistical information storage unit inputs an initial reference value obtained from statistical data obtained by actually measuring the processing performance information by applying a load of the input / output command and a normal value Criteria for determining a prediction reference value obtained by adding the processing performance information to the statistical data when the input / output command processing is executed on business Answer time determining means; initializing means for creating an initial table for obtaining the initial reference value; limit I / O processing number determining means for obtaining a performance limit of the number of input / output commands from the initial table; A disk array control unit comprising statistical data adding means for adding statistical data to a table.

16. The disk array control unit according to claim 15, wherein the processing performance information includes the number of input / output commands per unit time and a response time with respect to the number of input / output commands.

The disk array control unit according to claim 15 or 16, wherein the performance determination unit performs the determination by comparing the prediction reference value and the processing performance information.

The initial reference value is a straight line connecting the response time at the start of change of response time with respect to the number of input / output commands and after the start of change, and the input / output command at a pre-performance limit with a margin for the performance limit. And the prediction reference value is the response time at the number of input / output commands closest to the pre-performance limit in the processing performance information added to the statistical data and after the start of the change. The disk array according to any one of claims 15 to 17, wherein the response time is an intersection of a straight line connecting the response times of the data and the number of input / output commands of the pre-performance limit. Control unit.