About this article: This article is reviewed against the cited sources. This is source research, not lab validation. Check the stated versions and your own requirements before making changes.
Asking for more disk space on a hosted server is usually simple when snapshots are out of the way: the provider enlarges the virtual disk and the Windows volume is extended into the new space. Going the other way is not. Broadcom states plainly that there is no supported way to shrink a VMware virtual disk once it has been provisioned. So request growth in measured steps, and lay out each server's disks so the volume that will grow sits last on its own disk.
Why it matters
Where a hosting plan charges for the space allocated, over-asking "to be safe" means paying for space you may never use and cannot easily hand back. Under-planning the layout means the next request for space turns into a weekend job instead of a five-minute change. Both are avoidable with a little thought at build time.
How it works
Growing: two layers, and the condition that matters
Space has to be added at two layers: the virtual disk on the hosting platform, which the provider enlarges, and the volume inside Windows. Our companion article on which changes need a restart covers the platform conditions (snapshots committed first, some guests needing a reboot to see the new size, and the planned window database volumes need). This article is about the layout inside the disk.
Microsoft describes the Windows step as "extending the volume into available unallocated (empty) space on a disk drive", with an important condition: "The unallocated space must be located immediately after the volume that you want to extend." When the provider enlarges a disk, the new space appears at the end. If the volume you want to grow is the last one on that disk, extending is straightforward. If another partition, such as a recovery partition, sits after it, the new space is not adjacent and the job becomes harder.
Shrinking: inside Windows is possible, the virtual disk is not supported
Windows can shrink a volume, but with limits. Microsoft notes that files "like the paging file or the shadow copy storage area can't be automatically relocated" and "you can't decrease the allocated space beyond the point where the unmovable files are located".
Even after shrinking the volume inside Windows, the virtual disk on the platform stays the same size. Broadcom's article on growing, thinning and shrinking disks in ESXi is direct: "There is no supported way to shrink a virtual disk (vmdk file) after it has been provisioned." The workaround it describes uses a conversion tool and "will create a completely new virtual machine with a smaller disk", and it warns: "An effective backup strategy should always be used before performing permanent changes to existing virtual disks." In practice that is a planned project with downtime and a fresh test, not a routine request; ask your provider which tool and method it would use.
Partition style sets a ceiling
Every disk has a partition style. Microsoft's guidance says GPT "allows for volumes larger than 2 terabytes (TB)", and that "The default partition style is GUID Partition Table (GPT)" in Disk Management. A disk created long ago with the older MBR style, perhaps migrated from an old physical server, may hit a size ceiling that GPT would not.
Converting a data disk is not an in-place change. Microsoft says "You can convert a disk's partition style only if it doesn't contain any partitions or volumes", so the practical route is a new GPT disk and moving the data across. Microsoft's MBR2GPT tool converts "without modifying or deleting data on the disk", but "The tool can't be used to convert non-system disks from MBR to GPT". Check the partition style before a large growth request.
What to choose and why
Lay out disks for growth
Our editorial advice for a hosted Windows server:
- Keep the operating system on its own disk. It grows slowly and predictably.
- Put each data set that grows on its own virtual disk: file shares on one, database files on another, database logs on a third if the database is busy.
- Make the growing volume the only, or last, volume on its disk, so new space is always adjacent.
- Use GPT for data disks, so size ceilings are not a surprise later.
Ask for space in steps
Request enough for the coming planning period plus headroom, then grow again when measurements show the need. Growth is easy; reversing an oversized disk is not.
Before requesting more space
- Which disk and volume is filling, and how fast?
- Is the volume last on its disk, and is the disk GPT?
- Could old data be archived or cleaned up instead?
- Does the provider need snapshots removed first, or a time window?
A hypothetical example
A medical practice's server has one virtual disk laid out in this order: the system partition, then the partition holding the patient-document share, then a recovery partition at the end of the disk. The share fills up. The provider enlarges the disk, but the new space lands after the recovery partition, not directly after the share's partition, so the share cannot simply be extended. The fix is to add a new GPT data disk, move the share onto it in a quiet period, and from then on grow that disk alone. The example is illustrative.
Read next
Related articles you may want next.
- HostingAdding CPU, memory or disk to a hosted server: which changes need a restartPlan CPU, memory and disk upgrades for hosted servers. Check which changes support live expansion, the conditions that apply and when a planned restart is needed.Read article
- TroubleshootingWindows Disk Full, VMware Space Free: Map the LayersMap Windows volumes to virtual disks and datastores, then identify where capacity is missing before approving an expansion or filesystem cleanup.Read article
- HostingThin or thick provisioned storage: what you pay for and what can run outWhat thin and thick provisioning mean on a hosted VMware server, why a full shared datastore can stop a server whatever its disk format, and the questions to ask about billing and monitoring.Read article
Risks and limits
- Over-allocation can cost money for as long as the space is held, depending on how your plan charges.
- Shrinking a virtual disk is unsupported by VMware. The workaround builds a new virtual machine; plan it with a full backup and a tested result.
- Unmovable files limit Windows shrinks. The paging file and shadow copy storage pin the end of a volume.
- Old MBR data disks have a ceiling. Converting one means emptying it first; plan a new disk with your provider.
- Platform conditions apply. Snapshots and replication can delay disk growth.
Sources
Opened 15 November 2026; Broadcom and the two partition-style pages opened or reopened 16 November 2026.
- Microsoft Learn: Extend a basic or dynamic volume, page shows last updated 11 June 2025.
- Microsoft Learn: Shrink a basic volume, page shows last updated 25 April 2025.
- Microsoft Learn: Initialize new disks, page shows last updated 26 June 2025.
- Microsoft Learn: Convert a disk to GPT or MBR, page shows last updated 17 June 2025.
- Microsoft Learn: MBR2GPT, page shows last updated 14 July 2026.
- Broadcom Knowledge: Growing, thinning, and shrinking virtual disks in ESXi, article 341657 (page metadata: modified 17 September 2026).
Related article: Windows Disk Full, VMware Space Free: Map the Layers explains how the platform and Windows views of space differ.
Related service: Cloud hosting.
Learning notes
- Growth needs two steps: enlarge the virtual disk, then extend the Windows volume.
- A volume can only extend into space directly after it.
- VMware has no supported way to shrink a virtual disk; the workaround builds a new machine.
- GPT allows volumes over 2 TB; an MBR data disk must be empty to convert.
- Give each growing data set its own disk, with its volume last.
This article was originally published on .