Why Copying or Zipping Thousands of Small Cloud Files Is So Slow

Quick Answer

Copying thousands of small files from OneDrive, Google Drive, Dropbox, Proton Drive or Apple iCloud Drive can take much longer than copying a single large file of the same total size. The main reason is that each file requires individual processing, which may involve retrieving metadata, checking permissions, requesting or downloading the file, creating a destination entry, updating timestamps and attributes, and performing synchronization or security checks.

If the files are stored in the cloud and represented locally by placeholders, creating a ZIP archive can also be slow. The compression program may need to retrieve the actual contents of every file before it can add them to the archive. ZIP compression reduces the size of the data only when the selected compression method can compress it; it does not eliminate the work required to access thousands of individual files.

The most effective solution is usually to make the files available locally before copying or compressing them, where practical. If the source contains a very large number of small files, downloading an existing archive or creating an archive on a system that already has the files locally can be substantially faster than processing every cloud-backed file individually.

Why Thousands of Small Files Take So Long to Copy

File transfer speed depends on more than the total size of the data. The number of files, file sizes, storage performance, network latency, cloud synchronization behavior and destination file system can all affect the result.

Consider two folders containing 10 GB of data:

• Folder A: One file measuring approximately 10 GB.
• Folder B: 100,000 files averaging approximately 100 KB each.

Both folders contain approximately the same amount of data, but Folder B may take considerably longer to copy. Every file introduces additional work, even when its contents are small.

1. Per-file processing overhead

Operating systems cannot treat an ordinary folder copy as one continuous data transfer. They must enumerate the directory and create or update a destination entry for each file. Depending on the operation and file system, additional metadata, attributes, timestamps, security information and directory entries must also be processed.

For example, copying 100,000 files requires substantially more file-management operations than copying 100 files of the same combined size. These operations consume CPU time and storage input/output operations, even when each individual file takes only a fraction of a second to process.

2. Network latency and individual requests

Cloud storage applications communicate with remote services over a network. Some operations require metadata requests, file-content retrieval, authentication checks or synchronization coordination. These operations may be grouped, cached or performed concurrently, depending on the application and its implementation.

When a workload involves many small files, the overhead associated with individual operations can become a significant part of the total transfer time. A fast broadband connection does not necessarily eliminate this bottleneck because bandwidth and latency are different factors.

Bandwidth determines how much data can be transferred per second. Latency is the time required for a request and response to travel between systems. A connection with high bandwidth but relatively high latency can perform well when transferring large files while performing less efficiently on workloads containing thousands of small files.

3. Cloud placeholders and on-demand downloads

Some cloud storage applications show files in File Explorer or Finder without storing their complete contents on the local disk. These entries allow users to browse cloud files without downloading everything.

When a program attempts to read an online-only file, the cloud application may need to download its contents first. If a ZIP utility encounters thousands of such files, it can trigger a large number of retrieval operations. The application may also need to wait for each file to become available before continuing.

This behavior is particularly important when compressing a cloud-backed folder. The files may appear to be present locally, but their names and metadata being visible does not necessarily mean their contents are available offline.

4. Destination storage and file system overhead

The destination must create a separate file-system entry for every copied file. A destination with slow random input/output performance, limited free space or a busy storage device can increase the time required to complete these operations.

Mechanical hard drives are generally more sensitive to workloads involving many small, scattered operations than modern SSDs. However, an SSD cannot remove cloud-download delays, application overhead or network latency.

5. Antivirus and security scanning

Security software may inspect files as they are created, accessed or modified. This can add overhead during a large copy or extraction operation, particularly when thousands of individual files are involved.

The effect varies by security product, file type, system policy and workload. Do not assume antivirus software is the cause merely because copying is slow, and do not disable security protection as a first troubleshooting step.

Why Creating a ZIP File Can Also Be Slow

ZIP compression combines multiple files into an archive. Depending on the compression method and settings, it may also reduce the amount of data stored in that archive. However, creating a ZIP file is a separate processing operation, not an automatic shortcut around cloud downloads or file-system overhead.

1. Every file must be opened and read

To create a conventional ZIP archive containing actual file contents, the compression application must enumerate the source files, read their data, process the contents and write the archive entries. It must also maintain archive metadata, including file names and other information needed to extract the files later.

If the source files are online-only, the application may have to download each file before it can read it. As a result, the time required to create the archive may be dominated by cloud retrieval rather than compression itself.

2. Compression is CPU-intensive

Compression algorithms examine data and attempt to represent it more efficiently. The amount of CPU time required depends on the selected algorithm, compression level, file types, processor performance and implementation.

Increasing the compression level may improve the archive's size for some files, but it can also increase processing time. The smallest possible archive is not necessarily the fastest archive to create.

3. Some file types compress very little

JPEG photographs, MP4 videos, many modern image formats, encrypted files and files that have already been compressed often offer limited additional size reduction.

Creating a ZIP archive from these files can still be useful for packaging many files into one transferable object, even if the resulting archive is only slightly smaller than the original data.

4. Thousands of tiny files create archive overhead

Each file added to an archive requires its own entry and associated metadata. Even when individual files contain very little data, the compression program must still manage those entries.

For a folder containing thousands of small text documents, configuration files or source-code files, the per-file processing overhead may be more important than the amount of data being compressed.

5. The archive itself may need to synchronize

If you create a ZIP file inside a folder managed by OneDrive, Google Drive for desktop, Dropbox, Proton Drive or iCloud Drive, the cloud application may detect the new archive and upload it afterward.

The local compression operation and the cloud upload are separate stages. A ZIP utility may finish creating the archive while the cloud client is still uploading it. In other cases, background synchronization can compete with the compression operation for CPU, storage or network resources.

Understanding the Technology Behind Cloud File Transfers

Cloud storage is not simply a remote hard drive connected through the internet. Desktop clients combine local file-system integration, cloud APIs, synchronization engines, caching, authentication and background transfer mechanisms.

File metadata versus file contents

Metadata describes a file without necessarily containing its actual data. It can include the file name, size, modification time, identifier, sharing information and synchronization status.

A cloud application may retrieve or display this information without downloading the complete file. This makes browsing large collections more convenient, but a bulk copy or compression operation still needs the actual contents of the files being transferred or archived.

Files On-Demand and virtualized files

Cloud clients can integrate with the operating system so that cloud-backed files appear in normal folders. The operating system and client coordinate file access and retrieval when the contents are not stored locally.

These features reduce local storage requirements, but workloads that read every file in a directory can trigger large numbers of downloads. This is one reason a folder can open quickly in File Explorer while copying or compressing its contents takes much longer.

Concurrency and parallel processing

Cloud clients and archive utilities may process several files at the same time. Parallel processing can improve performance by overlapping network requests, disk operations and CPU work.

However, increasing concurrency does not guarantee faster results. Excessive simultaneous requests can increase resource contention, cause throttling or overload a slow destination. The appropriate behavior depends on the client, service limits, network and system resources.

Why one large file can transfer faster

A single large file can often be transferred as a more continuous stream of data. Once the transfer is underway, the system can spend a larger proportion of its time moving actual file contents instead of repeatedly creating file entries and managing separate requests.

This is why transferring one 5 GB archive may be much faster than transferring 50,000 individual files whose combined size is also 5 GB. The comparison is not absolute: the result depends on compression, network conditions, file sizes and the applications involved.

How OneDrive, Google Drive, Dropbox, Proton Drive and iCloud Differ

The general small-file problem affects many cloud storage services, but their local storage and synchronization models are not identical. The exact behavior depends on the operating system, application version, configuration and whether the files are already stored locally.

Service	Relevant technology or feature	What it means for copying and ZIP creation
Microsoft OneDrive	Files On-Demand	Online-only files can appear in File Explorer without their complete contents being stored locally. Reading or copying them out of the OneDrive location may require downloads.
Google Drive	Stream files and Mirror files	Streaming primarily keeps files in the cloud and makes them available on demand. Mirroring keeps a local copy of My Drive files, which can help avoid repeated cloud retrieval during local processing.
Dropbox	Online-only and available offline files	Online-only files use placeholders and download when accessed. Making a source folder available offline before a bulk operation can avoid triggering downloads one by one during compression.
Proton Drive	On-demand access and desktop synchronization	The Windows application supports cloud-backed files that are downloaded when accessed. Availability and behavior depend on the application and configured synchronization folders.
Apple iCloud Drive	Optimize Mac Storage	macOS can manage local storage by retaining some files in iCloud and downloading them when needed. Accessing many files may therefore involve additional downloads.

Microsoft OneDrive

OneDrive Files On-Demand lets users browse cloud files in File Explorer without downloading all their contents. Files marked as always available are downloaded and retained locally, while online-only files can be downloaded when accessed.

For a large bulk operation, right-click the relevant file or folder and select Always keep on this device. Allow the download to finish before copying or compressing the files. After confirming that the local operation is complete, you can use Free up space if you want to remove the local copies while retaining them in OneDrive.

Microsoft also documents the possibility of OneDrive remaining on “Processing changes” when many files have been added. This does not automatically mean the client is broken.

Google Drive for desktop

Google Drive for desktop offers streaming and mirroring for My Drive files. Streaming primarily accesses files from the cloud, whereas mirroring maintains local copies as well as cloud copies.

If you regularly process large collections of small files, mirroring may be useful when local disk capacity permits. It can reduce the need to download each file when a local application reads it. It does not eliminate the time needed to copy the files to another destination or to create the archive.

Before switching modes, confirm that files have finished syncing and that you understand where the local files are stored. Follow Google's instructions rather than manually deleting synchronization folders during a mode change.

Dropbox

Dropbox distinguishes between online-only files and files available offline. An online-only file can appear in the Dropbox folder without occupying the space required for its full contents. Opening it triggers a download.

For a large ZIP operation, use the Dropbox desktop application's Make available offline option on the source folder when sufficient local storage is available. Wait for the downloads to finish before starting compression.

Proton Drive

The Proton Drive Windows application supports synchronizing selected local folders and accessing cloud files on demand. Files stored in its cloud-backed folders may need to be downloaded when an application attempts to open them.

For a large local archive operation, ensure that the source contents are actually downloaded and accessible offline before starting. Check the current Proton Drive application documentation for the available synchronization controls on your operating system.

Apple iCloud Drive

iCloud Drive integrates with Finder and can work with the macOS Optimize Mac Storage feature. When local storage needs to be managed, macOS may remove locally stored copies of some files while retaining them in iCloud.

Before compressing a large collection, ensure that the files are available locally. The exact controls for downloading or retaining files depend on the macOS version and file status. Keep sufficient free disk space for both the downloaded source files and the resulting archive.

Recommended Solution: Prepare the Files Before Copying or Zipping

The most reliable general approach is to separate cloud retrieval from local file processing. First ensure that the source files are available locally, then perform the copy or compression operation.

Step 1: Check whether the files are online-only

Open the source folder in File Explorer or Finder. Look for the cloud provider's status icons or availability information.

• If the files are online-only, they may need to be downloaded before they can be processed locally.
• If the files are already stored locally, slow performance is more likely to involve file-system overhead, storage performance, security scanning, CPU use or another bottleneck.
• If the folder is still synchronizing, allow synchronization to finish before running a large operation where practical.

Step 2: Make the source available offline

Use the cloud application's supported option to download or retain the required files locally. The option's name varies by service.

• OneDrive: Select Always keep on this device.
• Google Drive: Consider mirroring My Drive files if appropriate for your workflow and available disk space.
• Dropbox: Select Make available offline.
• Proton Drive: Use the supported synchronization and download behavior for the Windows application or your relevant client.
• iCloud Drive: Ensure that the required files are downloaded and retained locally using the available Finder and macOS controls.

Wait until the files are fully downloaded. A visible file name or folder listing alone is not proof that every file's contents are stored locally.

Step 3: Check available storage space

Downloading the source files and creating an archive can temporarily require more disk space than the source folder alone occupies. The archive may also be large if the source data is already compressed or cannot be compressed effectively.

Ensure that the source files, destination copy and archive can fit on the relevant storage devices. Do not delete the original files to make room until you have verified the archive and confirmed that another usable copy exists.

Step 4: Copy or compress the local files

Once the files are available locally, use File Explorer, Finder or a trusted archive utility to copy or compress them. Choose a destination with sufficient free space and good performance.

If your goal is simply to transfer a folder to another computer, creating a ZIP archive may help by consolidating many files into one object. If your goal is to reduce the amount of data transferred, the benefit depends on how compressible the files are.

Step 5: Verify the result

After copying, compare the source and destination file counts and total sizes where practical. For important data, use a trusted file-comparison or checksum tool to verify file integrity.

After creating a ZIP archive, confirm that the archive opens successfully and that its expected files are present. For critical data, test the archive or extract it to a separate temporary folder and inspect the extracted contents.

What If the Files Are Already Stored Locally?

If the entire source folder is already downloaded but copying or compression remains slow, investigate the local workload rather than repeatedly changing cloud settings.

1. Compare a small sample

Copy a representative sample of files to a local folder outside the cloud-synchronized directory, then repeat the operation. Do not move the original files or delete the cloud copy.

If the local copy is substantially faster, the cloud client or synchronization activity may be contributing to the delay. If it is still slow, investigate storage performance, security scanning, file-system overhead and the archive utility.

2. Monitor CPU, disk and network activity

On Windows, open Task Manager and examine the Processes and Performance tabs while the operation runs.

• High network activity: The cloud client may still be downloading or uploading files.
• High disk activity: Storage operations or security scanning may be limiting throughput.
• High CPU activity: Compression or another processing task may be the bottleneck.
• Low activity across the system: The operation may be waiting on network responses, file-system operations or an application-specific limitation.

These observations are clues, not definitive diagnoses. Multiple bottlenecks can occur at the same time.

3. Check for synchronization activity

Look at the cloud application's status and activity window. If it is still processing many changes, wait for the queue to settle where practical. Avoid repeatedly pausing, resuming or restarting the client without a clear reason.

4. Check the destination drive

Confirm that the destination has enough free space and is not experiencing errors or unusually high activity. Copying many small files to a network share, external hard drive or slower USB device may take longer than copying them to an internal SSD.

5. Consider security scanning without disabling protection

If disk activity suggests that security scanning is contributing to the delay, review the security product's documented performance guidance and relevant logs. Do not create broad exclusions or disable real-time protection merely to speed up a copy operation.

Should You ZIP the Files Before Downloading Them?

Sometimes, but only when the workflow supports it. There is an important distinction between downloading an existing archive and creating a new archive from individual cloud files.

Situation	Recommended approach	Reason
An archive already exists in the cloud	Download the archive directly	One archive transfer can avoid handling every extracted file individually during download.
The source contains thousands of online-only files	Download the files first, then create the archive locally	This separates cloud retrieval from compression and makes the local processing stage easier to diagnose.
The source folder is already local	Create a ZIP archive locally if useful	The compression utility can read the local files without requiring cloud retrieval for each file.
The cloud provider supports creating an archive server-side	Consider using the provider's supported download or export option	Server-side packaging can avoid performing all file-reading and compression work on the local computer, when the service actually offers that capability.
The files are already compressed videos, photographs or archives	Use ZIP mainly for packaging, if needed	Additional compression may provide little size reduction while still consuming processing time.

Important: Do not assume that OneDrive, Google Drive, Dropbox, Proton Drive or iCloud Drive provides a server-side ZIP operation for every folder or account type. Available web download and export behavior varies by service, file selection and account configuration.

Can You Use 7-Zip or Another Archive Utility to Make It Faster?

A different archive utility may perform better for a particular workload because implementations differ in threading, buffering, compression algorithms and file handling. However, changing the compression program cannot eliminate the need to retrieve cloud-only file contents.

For locally stored files, a trusted utility such as 7-Zip may provide useful control over compression level, archive format and processing behavior. Test it with a representative sample before committing to a very large archive.

For maximum speed, choose a suitable low-compression or store mode if your utility supports it and a smaller archive is not essential. The exact setting and performance depend on the utility and data. A ZIP archive created without compression still requires reading the source files and writing archive entries, so it will not eliminate per-file overhead.

Can You Improve Speed by Splitting the Folder into Smaller ZIP Files?

Splitting a very large collection into multiple archives can make individual jobs easier to manage and may allow separate jobs to run independently. However, it is not automatically faster overall.

Creating several archives can repeat directory enumeration and file-handling work. Running too many compression jobs simultaneously can also increase disk contention, CPU usage and cloud-client activity.

Consider splitting the collection when you need manageable archive sizes, easier recovery after an interrupted transfer or separate logical groups. For the fastest result, benchmark one archive against a small number of separate archives using representative local files.

Common Mistakes to Avoid

• Compressing online-only files immediately: The archive utility may trigger downloads for thousands of files.
• Assuming file size alone determines speed: File count, latency, metadata operations and destination performance also matter.
• Increasing compression to the maximum: This may reduce the archive size but can increase CPU processing time.
• Creating the ZIP inside a busy synchronized folder: The cloud client may begin uploading the archive while other synchronization work continues.
• Changing cloud synchronization modes without checking status: An incomplete synchronization or incorrect deletion can result in missing local copies or data loss.
• Disabling antivirus protection as a first step: This introduces security risk without proving that security scanning is the cause.
• Deleting the source immediately after compression: A ZIP file should be verified before the source is removed.

Frequently Asked Questions

Why does copying 10,000 small files take longer than copying one large file?

Each small file requires file-system and application-level processing, including directory handling, destination entry creation and potentially separate cloud retrieval operations. A single large file generally has less per-file overhead relative to the amount of data transferred.

Why is creating a ZIP file from OneDrive or Google Drive so slow?

If the files are online-only or streamed, the archive utility may need to download the contents before it can compress them. Even when the files are local, processing thousands of entries can take time because of file-system overhead, compression work and security scanning.

Does ZIP compression make file copying faster?

It can reduce transfer time when the archive is substantially smaller than the original collection and the time saved during transfer exceeds the time needed to create and extract it. It can also simplify transfers by combining many files into one archive. If the data is already compressed or archive creation is slow, the total workflow may not be faster.

Will an SSD solve the problem?

An SSD can improve local storage performance, especially for workloads involving many small files. It cannot eliminate network latency, cloud download delays, synchronization overhead or CPU-intensive compression. If the bottleneck is cloud retrieval, upgrading the local drive alone may have little effect.

Is it better to download the files through a browser or a desktop sync application?

Neither method is universally faster. Browser downloads may package selected files into an archive or download files individually, depending on the service and selection. Desktop applications may access files through synchronization folders or on-demand mechanisms. Compare the available methods using a representative sample and account for whether the data is already local.

Why does the copy speed show zero bytes per second for a while?

The application may be waiting for a file to download, processing metadata, scanning files, creating destination entries or waiting for storage operations. Check the cloud client's status and Task Manager activity to identify which component is active. A temporary zero reading does not, by itself, prove that the operation has failed.

Does increasing internet speed always fix slow small-file transfers?

No. A faster connection can help when available bandwidth is the limiting factor. It may have a smaller effect when the bottleneck is latency, per-file processing, cloud-service limits, local storage or compression overhead.

Can I safely delete local cloud files after making a ZIP archive?

Only after verifying the archive and confirming that the original data remains safely stored elsewhere. Deleting files inside a synchronized cloud folder can propagate the deletion to the cloud and other synchronized devices. Use the provider's documented space-management feature when your intention is to remove only local copies.

What is the best method for transferring a folder containing hundreds of thousands of files?

Where practical, create or obtain an archive on a system that already has the complete source data locally, then transfer the archive and extract it at the destination. If the cloud provider supports an appropriate server-side export or packaging operation, evaluate that option. Test the workflow with a sample first and verify the resulting files.

Does a ZIP archive always take less space than the original files?

No. Compression effectiveness depends on the file contents and algorithm. Already-compressed or encrypted files may become only slightly smaller, and archive metadata can make the result marginally larger in some cases.

Conclusion

Slow copying and ZIP creation involving thousands of small cloud files are usually related to the combined cost of per-file processing, cloud retrieval, network latency, storage operations and compression. The total amount of data is only one part of the performance equation.

For the most predictable results, first confirm whether the files are stored locally, make cloud-backed files available offline when appropriate, allow downloads to finish, and then perform the copy or compression operation. Monitor CPU, disk and network activity if the problem continues. When possible, transferring an existing archive or creating one where the files are already local can avoid unnecessary individual cloud retrieval operations.

Sources

• Microsoft Support – Save disk space with OneDrive Files On-Demand for Windows
• Microsoft Support – Troubleshoot OneDrive sync issues stuck on Processing changes
• Microsoft Support – Zip and unzip files
• Google Drive Help – Stream and mirror files with Drive for desktop
• Dropbox Help – How to use Dropbox to save hard drive space
• Proton Support – How to sync folders with the Proton Drive Windows app
• Apple Support – Set up iCloud Drive