Docs Menu
Docs Home
/ /

AIクラスター 支援

AIクラスター アドバイザーは、 Atlas UIに埋め込まれたAI強化のヘルパーです。クラスターを作成したり、クラスターの構成を編集したりするときに使用して、クラスター構成に関する自然言語の質問に答えることができます。その後、 AIクラスター エージェントはワークロードとニーズに基づいて設定を推奨し、それらの設定をクラスターに直接適用できます。

Atlas でAIクラスター エージェントを使用するには、まず、Organization Settings ページで生成系AI機能を有効にしていることを確認します。

この組織レベルの設定を有効にするには、次の手順に従います。

1
  1. まだ表示されていない場合は、以下から目的の組織を選択しますナビゲーション バーのOrganizationsメニュー

  2. サイドバーで、Organization Settings をクリックします。

[ 組織設定]ページが表示されます。

2

Enable Atlas features that use generative AI トグルを On に設定します。

クラスターを作成 したり、 クラスターの構成を編集 したときに、Atlas UIでAIクラスター エージェントにアクセスできます。クラスターの作成中にAIクラスター 支援にアクセスするには、次の手順に従います。

1
  1. まだ表示されていない場合は、希望するプロジェクトを含む組織を選択しますナビゲーション バーのOrganizationsメニュー

  2. まだ表示されていない場合は、ナビゲーション バーのProjectsメニューから目的のプロジェクトを選択します。

  3. サイドバーで、 Database見出しの下のClustersをクリックします。

[ Clusters (クラスター) ] ページが表示されます。

2

すでに 1 つ以上のクラスターがある場合は、Create をクリックして Deploy your cluster ページを表示します。

これが最初のクラスターである場合は、Build a Cluster をクリックして Deploy your cluster ページを表示します。

3

ページの下部に移動し、 [Go to Advanced Configuration] をクリックします。

AIクラスター 支援のチャットウィンドウがページの右側に表示されます。

クラスターの編集中にAIクラスター 支援にアクセスするには、Upgrade <yourClusterName> ページで [Edit additional configuration] をクリックします。

支援を閉じるには、 チャットウィンドウの右上にある [] をクリックします。支援を再度開くには、[] をクリックします。パネルを閉じても、構成が変更されたり、そのセッションからチャット履歴が削除されたりすることはありません。同じセッション中にいつでもチャットウィンドウを再度開くことができます。

1

「 AIクラスター 支援へのアクセス 」の手順に従います。

2

クラスター構成に関するアドバイスを求めるプロンプトを送信します。ワークロード、ユースケース、または現在の問題点についてディスカッションすると、 AIクラスター アドバイザーがニーズに合わせて新しい構成を提案してください。

3

AIクラスター エージェントは、ニーズを説明し、クラスター構成を提案する応答を提供します。応答と推奨される構成を確認し、それらが満たされているかどうかを評価します。

4

必要に応じて、 AIクラスター エージェントに追跡プロンプトを送信して、推奨される構成の詳細情報を取得したり、ニーズに適した新しい構成をリクエストたりできます。

5

AIクラスター エージェントが提案する構成を受け入れるには、Apply these suggestions をクリックします。これは構成を確定せず、変更を続けることができます。

6

チャットウィンドウのAIクラスター エージェントの応答の下にある または ボタンを使用して、応答やAIクラスター エージェントの使用フィードバックに関するフィードバックを提供できます。

7

確認したら、クラスター構成を保存します。

クラスターを作成している場合は、ページの下部にある Create Cluster をクリックして、推奨される構成でクラスターを作成します。

クラスターを変更している場合は、ページの下部にある Review Changes をクリックして、変更の概要を表示します。次に、Accept Changes をクリックして新しい構成を保存します。

これにより、 AIクラスター 支援が閉じられます。

チャット中に、 AIクラスター 支援は次のタスクに応じて、異なる Atlas UIページへのリンクを提供します。 AIクラスター 支援は、次の Atlas UIページのいずれかにリンクする場合があります。

次のシナリオでは、Atlas でクラスターを作成または編集する際にAIクラスター 支援 を使用する方法の一部を示します。

ユースケースとニーズを説明し、 AIクラスターエージェントに質問して、どのクラスター構成が自分に最適であるかを提案してください。

例、アプリケーションを開発している場合は、 AIクラスター 支援にアプリケーションの目的と予想されるユーザー数を指示できます。

I’m setting up a development cluster for a restaurant rating app with
100 users in NYC. What’s the best cluster configuration?
For a small development app with ~100 users, cost-effectiveness and
reliability are key. I recommend starting with an M10 cluster on AWS in
US_EAST_1 (N. Virginia). This tier supports your use case with enough
resources for development and light testing, while keeping costs
controlled. Avoid free (M0) since you already have one and FLEX is
unnecessary for your scale.
I'll suggest a configuration based on these requirements. Quick
questions to further optimize:
- Do you expect your dataset to grow beyond 10GB during development?
- Any other environments needed, like staging or test?
- Will more developers or services connect to this cluster soon?
Suggesting the config now.

AIクラスター エージェントが構成を提案したら、それを受け入れて保存したり、満たす構成に達するまでニーズをディスカッションし続けます。

また、サポートが必要になると予想されるデータのサイズと、プロジェクトが基づいている場所をAIクラスター エージェントに指示することもできます。

I’m looking to migrate an existing project to Atlas. We currently
have 100GB of data that we need to move over. The project is currently
deployed in Europe and we want to maintain that. How should I configure
my cluster?
For migrating a 100GB project to Atlas in Europe, I recommend the following:
- Use AWS as your cloud provider (most commonly used, but let me know if
you prefer GCP or Azure).
- Select the region EU_WEST_1 (Ireland) to host your data in Europe.
- Choose at least an M40 instance (this tier is recommended for
migrating and running production workloads with 100GB+ data, ensuring
strong performance during data load and normal operations).
- Set storage to 150GB (100GB data + 30% headroom).
- Default disk IOPS for M40 will provide ample performance for both
migration and ongoing workload.
Let’s proceed with this setup.
A few quick clarifications:
- Is this a production environment with performance/uptime requirements?
- Do you expect rapid growth in data or traffic soon?
- Any preference for AWS, GCP, or Azure within Europe, or should we
proceed with AWS?
I’ll configure the cluster now with these optimized settings.

繰り返しになりますが、 AIクラスター 支援 を使用すると、提案された構成を受け入れて保存したり、ニーズに合った構成に作業を続けることができます。

ユースケースとニーズの変更を説明し、 AIクラスター アドバイザーがクラスター構成を変更するのに最適な方法を提案するようにします。

例、 AIクラスター 支援 と予算の制約について説明できます。

My cluster costs more than expected. Is there a setup that keeps
us under $500/mo?
To keep monthly costs under $500, I recommend switching to an M30 tier.
This provides reliable performance for standard production workloads at
around $394/month on AWS, plus some buffer for backups and storage.
Here’s how I’ll configure it:
- Instance size: M30 (entry-level dedicated, good for most production
apps)
- Region: N. Virginia (US_EAST_1) (stays consistent with your existing
clusters)
- Disk size: 40GB (default for M30, suitable for moderate workloads)
- No additional IOPS or premium storage (keeps costs down)
- No analytics/search node or advanced add-ons
Let’s proceed with this cost-effective configuration.
Before finalizing, can you clarify:
- Is this cluster for production, staging, or development?
- What’s your estimated data volume (just rough amount of GB needed)?
- Any specific performance requirements (high writes, reads, etc.)?
I’ll now suggest the configuration for you.

AIクラスター エージェントが新しい構成を提案します。

AIクラスター 支援 を使用して、増加したワークロードについて説明することもできます。

Suggest a cluster setup as I'm expanding to India and expect 20,000
new users and the associated data.
Since you're expanding to India with 20,000 expected users, you’ll need
a production-grade cluster deployed in an appropriate Indian region for
low latency. I recommend AWS as the cloud provider (unless you have a
preference for GCP or Azure), using the AP_SOUTH_1 region (Mumbai).
For your user base and workload size, an M30 dedicated tier is a strong
fit—it’s ideal for mid-sized production apps. This setup will handle
moderate to high traffic and provides headroom as your user base grows.
I'll set:
- Cloud Provider: AWS
- Region: AP_SOUTH_1 (Mumbai, India)
- Instance Size: M30 (production-ready with sufficient resources)
- Disk Size: Default (40GB, upgrade later if data grows quickly)
To refine this setup:
- What type of workload will the cluster handle (e.g., read-heavy,
write-heavy, analytics)?
- Any data volume estimate (GB/TB) for the first year?
- Any specific cost constraints?
Applying this recommendation now.

AIクラスター エージェントは、スケーリングされたニーズに対応できる構成を提案します。

  • AIクラスター 支援のデータ使用量

戻る

クラスターの作成

項目一覧