Teamprise Remote Accelerator Benchmarks
Teamprise Remote Accelerator acts as a local version control proxy for Team Foundation Server, to provide blazing fast access to version control data for remote developers.
- Developer 1: Working from a home office in Chicago, IL
- Developer 2: Working from a home office in Randalstown, N. Ireland
- Developer 3: Working from a coffeeshop in Chicago, IL
- Developer 4: Working from a home office in Kochi, India
Teamprise has collected benchmark data from a number of Teamprise employees working in branch offices across the world, connecting to a shared Team Foundation Server in our headquarters.
Benchmarks show a considerable improvement in speed compared to connecting directly to Team Foundation Server without a version control proxy. One developer reported over 400% speed improvement over a direct connection to the server.
Such dramatic speed improvements are available thanks to the innovative "cache seeding" available in Teamprise Remote Accelerator, where Team Foundation Server resources are downloaded automatically to the client and cached immediately. New resources are then available immediately to the client.
| Developer 1 | Developer 2 | Developer 3 | Developer 4 | |
|---|---|---|---|---|
| Location | Chicago, IL (Home Office) |
Randalstown, N. Ireland (Home Office) |
Chicago, IL (Coffeeshop) |
Kochi, India (Home Office) |
|
Average Throughput (Direct to TFS) |
166 Kb/s | 113 Kb/s | 83 Kb/s | 61 Kb/s |
|
Average Throughput (Remote Accelerator) |
414 Kb/s | 459 Kb/s | 324 Kb/s | 227 Kb/s |
|
Improvement Using Remote Accelerator |
250% | 400% | 390% | 370% |
Our benchmark data attempts to replicate the real-life version control needs of software developers, however speed gains from Teamprise Remote Accelerator may vary on a number of factors, including speed of the client's connection to the Team Foundation Server, speed of the computer running Remote Accelerator, and size of file(s) being downloaded.
Methodology
Four Teamprise developers downloaded a number of files both directly
from a Team Foundation Server at Teamprise Headquarters in Champaign, IL,
as well as using Teamprise Remote Accelerator on their local computer.
Developer 1 performed this test from his home office in Chicago, IL, using a computer running Windows Vista (32-bit).
Developer 2 performed this test from his home office outside of Randalstown, Northern Ireland, using a computer running Windows Vista (64-bit).
Developer 3 performed this test from a coffeeshop in Chicago, IL, using a virtual machine running Windows XP SP3.
Developer 4 performed this test from his home office in Kochi, India, using a computer running Windows XP SP3.
All developers connect to the Team Foundation Server over an SSL-based VPN which both encrypts and compresses all data.
For each test, a total of 8.8 MB was downloaded, comprised of files ranging from 2 KB to 10 KB. Each test was executed using Microsoft's Team Foundation Server Command Line Client (tf.exe), and a total of four connections to the Team Foundation Server were made for each test. Tests compare downloading directly from Team Foundation Server to downloading using a seeded Remote Accelerator, such that files are delivered directly from the version control client cache.
Multiple tests were run by each developer, with the slowest and fastest results removed to normalize the data, and the remaining test data was averaged to produce the results.
Throughput: Direct to Team Foundation Server vs. Using Remote Accelerator

The darker bars show the throughput to clients using Remote Accelerator. The lighter bars show the throughput to clients direct to the Team Foundation Server.
Significant improvements were provided when using Remote Accelerator, thanks to its intelligent cache seeding.
Benchmarks showed a minimum of 250% improvement to the slowest developer's computer and 400% improvement to the fastest developer's computer.
The most significant improvements occurred for developers with relatively fast computers and moderate (1 Mb/s or higher) connections to the Internet.
The least significant improvements occurred for developers who had computers with slow I/O (for example, the Virtual Machine setup used by Developer 3) and developers with very slow connections to the Team Foundation Server. (When performing any version control operation, the client must still connect to the Team Foundation Server.)
Still more improvement is available when downloading large files, however larger files were excluded from these benchmarks to ensure that they reflected real-life scenarios as much as possible.

