skip to Main Content

Exchange database best practices

Before installing Exchange Server, it’s good to know the Exchange database best practices. Design first, and from there on, configure a great Exchange Server configuration. For example, how many databases should you configure on a volume? Can you keep the database and log folder on the same disk? In this article, you will learn how how to design Exchange database best practices.

Exchange Server architecture

There are two Exchange architectures:

  • Standalone – One single Exchange Server
  • High Availability (DAG) – Multiple Exchange Servers

Note: We recommend installing Exchange Server in High Availability. If one Exchange Server stops working for any reason, the mail flow will keep going. In addition, it will give you time to fix the broken Exchange Server without the users notifying problems.

Standalone Exchange Server

  • Single database per volume is recommended
  • Single log per volume is recommended
  • Both the same database and log (co-location) on a volume is not supported
  • Database size of 200GB per database or less
  • Windows basic disk type
  • GPT partitions instead of MBR
  • Configure volumes as mount points and not drive letters is recommended
  • Configure volumes for database and log as ReFS or NTFS
  • Allocation unit size of 64KB for both database and log volumes
  • Start the naming convention of the database with DB01 and keep increasing the number

Note: For recoverability, move the database (.edb) file and logs from the same database to different volumes backed by different physical disks.

High Availability Exchange Servers

  • Single database per volume is supported, is not required
  • Single log per volume is supported, is not required
  • Both the same database and log (co-location) is supported
  • The number of databases per volume should equal the number of copies of each database
  • Database size of 2000GB per database or less
  • Windows basic disk type
  • GPT Partitions instead of MBR
  • Configure volumes as mount points and not drive letters is recommended
  • Configure volumes for database and log as ReFS or NTFS
  • Allocation unit size of 64KB for both database and log volumes
  • Start the naming convention of the database with DB01 and keep increasing the number

Note: GPT is a disk architecture that expands on the older master boot record (MBR) partitioning scheme. GPT maximum formatted partition size is 256 terabytes. MBR maximum formatted partition size is 2 terabytes.

Exchange database examples

A couple of examples that will show the best practices for Exchange databases in a standalone or high availability architecture:

Example 1. You want to configure a standalone Exchange Server in the organization:

  • Place the database on a separate volume and place the log on a separate volume
  • Don’t add more than one database per volume and one log per volume
  • Both volumes backed by different physical disks
Exchange database best practices standalone architecture

Example 2. Suppose you have a DAG configured with two Exchange Servers:

  • Place the database and log on to the same volume
  • No more than two databases and two logs (co-location) on the same volume
  • Both volumes backed by different physical disks
Exchange database best practices DAG architecture two servers

Example 3. Suppose you have a DAG configured with three Exchange Servers:

  • Place the database and log on to the same volume
  • No more than three databases and three logs (co-location) on the same volume
  • All three volumes backed by different physical disks
Exchange database best practices DAG architecture three servers

Example 4. The configuration in the diagram shows four servers:

  • All four servers have the same four databases hosted on a single disk per server
  • The number of database copies configured per volume should equal the number of copies of each database
Exchange database best practices DAG architecture four servers

We hope that the Exchange database best practices help you in designing the Exchange environment.

Keep reading: Exchange database size recommendations »

Conclusion

You learned the Exchange database best practices. It’s best to configure an Exchange Server in High Availability with a Database Availability Group (DAG). When a mailbox database goes offline, it will failover the mailbox database to the other Exchange Server.

Did you enjoy this article? You may also like Create mailbox database in Exchange Server. Don’t forget to follow us and share this article.

ALI TAJRAN

ALI TAJRAN

ALI TAJRAN is a passionate IT Architect, IT Consultant, and Microsoft Certified Trainer. He started Information Technology at a very young age, and his goal is to teach and inspire others. Read more »

This Post Has 9 Comments

  1. Let me start by saying your site is awesome!
    Now the question:
    Why do you “Configure volumes as mount points and not drive letters is recommended”?
    I mean mount points are cool, but can make life confusing. And if you’re using a mount point, do you mount it as a subdirectory on the C: volume? If not how do you recommend placing / accessing the mount point?

    Thanks,
    Graham

    1. Thank you, Graham. Glad to hear that.

      When you have a small organization, you can use drive letters. But, when you need to create more than 20+ mailbox databases, you don’t have that many drive letters. So that’s why you will choose mount points.

      I can’t know how many mailbox databases an organization will have. That’s why the recommended way is mount points. It will work for anyone.

      To work with mount points, you can create a folder on the (C:) drive and mount the volumes to that folder.

  2. Hi Ali,

    thanks for your sharing.
    I have 2 Ex2019 Servers and a DAG Configuration. What do you think about active-active modus and active-passive modus ? For the High Availability which one of them is the best practice ?
    Thank you for your answer.

  3. Hi Ali, indeed a very good explanation there. However, I still cant find the answer, with 2 node DAG, can we have 1 active, 2 copies and 1 lagged? Or we must stick to 3 node, 1 active, 1 copy and 1 lagged?

  4. very good

    small mistake in exemple 2, on the left part under ex01, for DB02+logs it should be passive instead active.

    thanks agaign for very good articles

  5. Looking at the 4 server in the dag diagram; I’ve set up my environment in a way where Exchange 1 and 2 are at site A and server 3 and 4 are at site B;

    EXDB1 is active on EX1 with a passive on EX4
    EXDB2 is active on EX3 with a passive EX2
    EXDB3 is active on EX4 with a passive on EX1
    EXDB4 is active on EX2 with a passive on EX3

    Both sites are appropriately backed up – Would you consider these enough copies or should I ideally aim for all 4 to be copied, baring in mind we have 4 TB of data?

Leave a Reply

Your email address will not be published.