Skip to content


Amazon S3 Tar Tool

Go Build status go-version API Reference Apache V2 License

s3tar is utility tool to create a tarball of existing objects in Amazon S3.

s3tar allows customers to group existing Amazon S3 objects into TAR files without having to download the files, unless using the --concat-in-memory flag (see below). This cli tool leverages existing Amazon S3 APIs to create the archives on Amazon S3 that can be later transitioned to any of the cold storage tiers. The files generated follow the tar file format and can be extracted with standard tar tools.

s3tar operates in two distinct modes, each tailored for specific use cases. The default method is designed for optimal performance with large objects, making it ideal for generating tarballs that predominantly consist of substantial data. In this mode, s3tar executes operations primarily through the Amazon S3 backend, eliminating the need to download the data.

Conversely, the concat-in-memory method is specifically optimized for small objects, facilitating the concatenation of hundreds of thousands or millions of objects. This approach involves downloading the data into the instance and conducting most operations in the system's memory. Each method comes with its unique pricing structures, which are explained in the dedicated pricing section.

Using the Multipart Uploads API, in particular UploadPartCopy API, we can copy existing objects into one object. This utility will create the intermediate TAR header files that go between each file and then concatenate all of the objects into a single tarball.


The tool follows the tar syntax for creation and extraction of tarballs with a few additions to support Amazon S3 operations.

flag description required
-c create yes, unless using -x
-x extract yes, unless using -c
-C destination to extract yes when using -x
-f file that will be generated or extracted: s3://bucket/prefix/file.tar yes
-t list files in archive no
--extended to use with -t to extend the output to filename,loc,length,etag no
-m manifest input no
--region aws region where the bucket is yes
-v, -vv, -vvv level of verbose no
--format Tar format PAX or GNU, default is PAX no
--endpointUrl specify an Amazon S3 endpoint no
--storage-class specify an Amazon S3 storage class, default is STANDARD no
--size-limit This will split the tar files into multiple tars no
--concat-in-memory Enables building the tarball in memory by downloading the data. (more details below) no

The syntax for creating and extracting tarballs remains similar to traditional tar tools:

   s3tar --region region [-c --create] | [-x --extract] [-v] -f s3://bucket/prefix/file.tar s3://bucket/prefix

Storage Class Options

Storage Class


To create a tarball s3://bucket/prefix/archive.tar from all the objects located under s3://bucket/files/

s3tar --region us-west-2 -cvf s3://bucket/prefix/archive.tar s3://bucket/files/
# this will create tarballs that are 1GB in size
s3tar --region us-west-2 --size-limit 1074000000 -cvf s3://bucket/archive.tar s3://bucket/files/
# outputs: 
# s3://bucket/archive.01.tar 
# s3://bucket/archive.02.tar 
# s3://bucket/archive.03.tar 

Manifest Input

The tool supports an input manifest -m. The manifest is a comma-separated-value (csv) file with bucket,key,content-length and an optional etag. Content-length is the size in bytes of the object. For example:

$ cat manifest.input.csv

$ s3tar --region us-west-2 -cvf s3://bucket/prefix/archive.tar -m /Users/bolyanko/manifest.input.csv

# the manifest file can be a local file or an object in Amazon S3

$ s3tar --region us-west-2 -cvf s3://bucket/prefix/archive.tar -m s3://bucket/prefix/manifest.input.csv

# The manifest can also contain the etag as a fourth column if its is known
$ cat manifest.input.csv

Large-Objects vs Small-Objects (In Memory)

The original design of s3tar prioritized the creation of tarballs for large objects. Previously, users were facing challenges by having to meticulously adjust various factors such as instance size, EBS/Instance Store, memory, and network bandwidth to build tarballs on EC2 Instances. Recognizing the need for a more efficient process, s3tar was developed to eliminate the necessity for users to download data, opting instead to leverage Amazon S3 MultiPart Objects.

As users increasingly employed s3tar for creating tarballs of small objects, a new feature has been introduced to facilitate the direct download of data and in-memory tarball construction. This enhancement significantly improves both performance and cost efficiency. To illustrate, building a tarball containing 1 million small objects now takes approximately 6 minutes on a c7g.4xlarge, compared to the previous version's 3-hour timeframe. With this modification, s3tar prioritizes GET operations, minimizing most PUT operations, as the majority of PUTs occur in RAM. This strategic shift substantially reduces the overall cost of tarball construction. For instance, the cost of building the same 1 million-object tarball is now approximately $0.45 (us-west-2), as opposed to the non in-memory version's cost of around $10. Users that are creating tarballs of extensive small objects, numbering in the hundreds of thousands or millions, are recommended to leverage the --concat-in-memory flag for enhanced efficiency and better pricing. At this time the in-memory version does not include a TOC. Users will have to download the tarball if they wish to extract the contents.

TOC & Extract

Tarballs created with this tool generate a Table of Contents (TOC). This TOC file is at the beginning of the archive and it contains a csv line per file with the name, byte location, content-length, Etag. This added functionality allows archives that are created this way to also be extracted without having to download the tar object.

You can extract a tarball from Amazon S3 into another Amazon S3 location with the following command:

s3tar --region us-west-2 -xvf s3://bucket/prefix/archive.tar -C s3://bucket/destination/

To extract a single file in a tar, or a prefix

s3tar --region us-west-2 -xvf s3://bucket/prefix/archive.tar -C s3://bucket/destination/ folder/image1.jpg 
# or a dir
s3tar --region us-west-2 -xvf s3://bucket/prefix/archive.tar -C s3://bucket/destination/ folder/ 


If you want to list the files in a tar

s3tar --region us-west-2 -tf s3://bucket/prefix/archive.tar 

Generating manifest files

We can generate manifest files to pass to s3tar with other tools. This will allow us to apply advanced filtering. For example, using the AWS CLI and jq we can create a file and filter the date with --query:


$ aws s3api list-objects --bucket $bucket --prefix $prefix \
--query "Contents[?LastModified >= '"$date_start"' && LastModified < '"$date_end"' && Size > \`0\`].{Key: Key,Size: Size}" \
| jq -r '.[] += {"bucket": "'$bucket'"}' \
| jq -r '["bucket","Key","Size"] as $cols | map(. as $row | $cols | map($row[.])) as $rows | $cols, $rows[] | @csv' > manifest.csv

Another option is to use a CSV generated by Amazon S3 Inventory. For example, the following command will generate a manifest file for objects LastModified between 2022-12-01 and 2023-01-01.

# Column 1 is the bucket
# Column 2 is the key
# Column 6 has the size
# Column 7 has the LastModified timestamp
# Column 8 is the ETag
awk -F',' 'BEGIN {OFS=","} {gsub(/"/,"",$6);gsub(/"/, "", $7); if($6>0 && $7>="2022-12-01 00:00:00" && $7<"2023-01-01 00:00:00") print $1,$2,$6,$8}' s3-inventory.csv > output.csv


The tool's performance is bound by the API calls limitations. The table below has a few tests with files of different sizes.

Number of Files Final archive size Average Object Size Creation Time Extraction Time Estimated Cost (us-west-2) - Standard
41,593 20 GB 512 KB 6m10s 3m11s $0.4159
124,779 61 GB 512 KB 18m24s 10m5s $1.2478
249,558 123 GB 512 KB 40m56s 21m42s $2.4956
499,116 246 GB 512 KB 1h34m 38m58s $4.9912
748,674 369 GB 512 KB 2h36m $7.48674
14,400 73 GB 70 MB 2m15s 1m20s $0.1440
69,121 3.75 TB 70 MB 1h11m30s 32m20s $0.6912

The application is configured to retry every Amazon S3 operation up to 10 times with a Max backoff time of 20 seconds. If you get a timeout error, try reducing the number of files.


A make file is included that helps building the application for darwin-arm64 linux-arm64 linux-amd64. Place the resulting s3tar binary in your PATH.

If you get the error dial tcp: lookup i/o timeout it means your network is restricting access to that domain. You can bypass the proxy by setting the following variable: export GOPROXY=direct

How the tool works

This tools utilizes Amazon S3 Multipart Upload (MPU). MPU allows you to upload a single object as a set of parts. Each part is a contiguous portion of the object's data. You can upload these object parts independently and in any order. After all parts of your object are uploaded, Amazon S3 assembles these parts and creates the object.

Multipart upload is a three-step process: You initiate the upload, you upload the object parts or copy from an existing Amazon S3 Object, and after you have all the parts, you complete the multipart upload. Upon receiving the complete multipart upload request, Amazon S3 constructs the object from all the parts, and you can then access the object just as you would any other object in your bucket. You can learn more about Multipart Upload on the MPU Overview

There are two Amazon S3 API Operations that allow adding data to a Multipart Upload. UploadPart and UploadPartCopy. This tool generates TAR header files and uses s3.UploadPart to upload the header data into a MPU, and then it uses s3.UploadPartCopy to copy your existing Amazon S3 Object into the newly created object.

Currently Multipart Uploads have a minimum requirement of 5MB per part and each part can go up to 5GiB. The total maximum MPU object size is 5TiB.

s3tar automatically detects the size of the objects it needs to tar. The total size of all the files must be greater than 5MB. If the individual files are smaller than the 5MB multipart limitation the tool will recursively concatenate groups of files into 10MB S3 objects. The tool generates an empty 5MB file (zeros) and everything gets appended to this file, on the last file of the group a CopySourceRange is performed removing the 5MB pad. As a last step the tool will merge all the objects together creating the final tar.

Group1 = remove5MB([(((((5MB File) + header1) + file1) + header2) + file2)...])
Group2 = remove5MB([(((((5MB File) + header1) + file1) + header2) + file2)...])
NewObject = Concat(Group1, Group2)

If the files being tar-ed are larger than 5MB then it will create pairs of (file + next header) and then merge. The first file will have a 5MB padding, this will be removed at the end:

NewS3Object = [(5MB Zeroes + tar_header1) + (S3 Existing Object 1) + tar_header2 + (S3 Existing Object 1) ... (EOF 2x512 blocks)]

Testing & Validation

We encourage the end-user to write validation workflows to verify the data has been properly tared. If objects being tared are smaller than 5GB, users can use Amazon S3 Batch Operations to generate checksums for the individual objects. After the creation of the tar, users can extract the data into a separate bucket/folder and run the same batch operations job on the new data and verify that the checksums match. To learn more about using checksums for data validation, along with some demos, please watch Get Started With Checksums in Amazon S3 for Data Integrity Checking.


It's important to understand that Amazon S3's API has costs associated with it. In particular PUT, COPY, POST are charged at a higher rate than GET. The traditional mode of generating tarballs heavily favors Amazon S3 PUT operations, while the in-memory mode favors GET operations. Because of this, pricing is substantially different between the two. Please refer to the Amazon S3 Pricing page for a breakdown of the API costs. You can also use the AWS Cost Calculator to help you price your operations.

Traditional Amazon S3 backend operations

The majority of requests performed by in this mode are COPY and PUT operations.

During the build process the tool uses Amazon S3 Standard to work on files. If you are aggregating 1,000 objects, then it will require at least 1,000 COPY operations and 1,000 PUT operations for the tar headers.

Example: If we want to aggregate 10,000 files

$0.005 PUT, COPY, POST, LIST requests (per 1,000 requests)
To Copy the 10,000 files to the archive we will do at least 10,000 COPY operations

10,000 / 1,000 * $0.005 = $0.05 
We need to generate at least 10,000 header files, that's 10,000 PUT opeartions

10,000 / 1,000 * $0.005 = $0.05 
There are other intermidiate operations of creating multipart
It would cost a little over $0.1 to create an archive of 10,000 files

The cost example above only prices the cost of performing the operation. It doesn't include how much it would cost to store the final object.

In-Memory Tarball Generation

This mode works by downloading (GET) the small files and building the tarball in memory. The majority of operations in this mode are GET operations. The formula to estimate the cost of building a tarball in this mode is as follows:

(Number of files * GET Pricing) + (number of MultiPart parts * PUT Pricing) 

MultiPart Objects are limited at 10,000 parts. The following example illustrates pricing for a tarball with 1M objects and 10,000 parts in us-west-2:

(1,000,000 * $0.0000004) + (10,000 * $0.000005) = $0.45

Limitations of the tool

This tool still has the same limitations of Multipart Object sizes:

  • The cumulative size of the TAR must be over 5MB
  • The final size cannot be larger than 5TB


See CONTRIBUTING for more information.

Frequently Asked Questions (FAQ)

Does the tool download any files?

No, all files are copied from their current location in Amazon S3 to their destination using the s3.UploadPartCopy API call.

Does the tool upload any files?

We are using the go archive/tar library to generate the Tar headers that go in between the files. These files are uploaded to Amazon S3 and concatenated with the Multipart Upload.

Does the tool delete any files?

No, the original files will remain untouched. The user is responsible for the lifecycle of the objects.

Is compression supported?

No, the tool is only copying existing data from Amazon S3 to another Amazon S3 location. To compress the objects it would require the tool to download the data, compress and then re-upload to Amazon S3.

Are Amazon S3 tags and meta-data copied to the tarball

No. Currently we're storing the Etag in the TOC, there is a possibility that could allow us to expand this.

What size of files are supported?

Any size that is within the Amazon S3 Multipart Object limitations. On the small side they can be as small as a few bytes, as long as the total archive at the end is over 5MB. On the large side the max size per object is 5GB, and the total archive is 5TB.

Can I open the resulting tar anywhere?

Yes, the tarballs are created with either PAX (default) or GNU headers. You can download the tar file generated and extract it using the same tools you use to operate on tar files.

What type of compute resources do I need to run the tool?

Since the tool is only doing API calls, any compute that can reach the Amazon S3 API should suffice. This could run on a t4g.nano or Lambda, as long as the number of files is low enough for the 15 minute window.


This project is licensed under the Apache-2.0 License.