Updated: 2009-04-23
This article provides general recommendations for bandwidth and latency ranges for Microsoft Office SharePoint Server solutions and includes test results for common operations based on specific bandwidth and latency combinations.
In this article:
A key consideration in determining which of the supported deployment solutions to implement for your Office SharePoint Server deployment is the available network bandwidth and the latencies between your sites. Latency is the time required for a packet to travel from one point on a network to another. The recommendations and test results included in this guidance are based on latency that is measured one way, not round-trip. Latency can influence planning for geographic deployments in which data travels across WAN links that span multiple cities, states, provinces, countries, or continents.
When measuring the latency that affects users, use the Ping tool (ping.exe) to measure latency from the client computer to the Web server on the server farm. Divide the round-trip result by two. When measuring the latency between two data centers that host servers of the same farm, use the Ping tool to measure latency from a Web server in the remote data center to the database server in the primary data center. Divide the round-trip result by two.
This article includes test results based on specific bandwidth and latency targets for the following:
The bandwidth statistics and recommendations within this guidance are based on test results produced in the Microsoft Technology Center (MTC) in Austin, Texas. Microsoft customers are welcome to use any of the MTC labs around the world or the Microsoft Enterprise Engineering Center in Redmond to extend this testing to their own deployment scenarios. For more information, see the following:
Important All test results described in this paper were generated in a simulated environment with no customizations. These test results represent the minimum response times that can be expected in a WAN environment. Applying customizations, utilization of multiple services and features, and additional networking infrastructure will increase the response times.
Recommended bandwidth ranges
This section provides generally recommended bandwidth ranges for each of the three solutions described in Supported global solutions for Office SharePoint Server (central, central with regional sites, and distributed). The recommendations are provided in the following table and are based on a user concurrency of 10 percent. For example, the recommendations for 10,000 users assume that only 1,000 users are requesting page views at the same time.
| Number of users | Central | Central with regional sites | Distributed |
|---|
|
100-5,000
|
Bandwidth ranges of 3 megabits per second (Mbps) (dual T1) and greater with latencies no greater than 100 milliseconds (ms)
|
Bandwidth ranges of 1.5 Mbps (T1) and greater with latencies no greater than 100 ms
|
Bandwidth ranges of 1.5 Mbps (T1) and greater with latencies no greater than 100 ms
|
|
10,000
|
Bandwidth ranges of 3 Mbps (dual T1) and greater with latencies no greater than 250 ms
|
Bandwidth ranges of 1.5 Mbps (T1) and greater with latencies no greater than 500 ms
|
Bandwidth ranges of 1.5 Mbps (T1) and greater with latencies no greater than 500 ms
|
|
100,000
|
Bandwidth ranges of 3 Mbps (dual T1) and greater with latencies no greater than 250 ms
|
Bandwidth ranges of 1.5 Mbps (T1) and greater with latencies no greater than 500 ms
|
Bandwidth ranges of 1.5 Mbps (T1) and greater with latencies no greater than 500 ms
|
Average page-load times for common user operations
Average page-load time is defined as the time, in seconds, for a request to display a Web page. It is measured from the time a page request is sent by a browser to the time that all bytes for the page and all dependent images and resources are received from the Web server.
Testing for this category was based on the following ratio of page requests:
| Page requested (read only) | Percentage of overall requests |
|---|
|
WSSHomePage
|
25
|
|
WSSSmallDocLibAllItems
|
14
|
|
WSSDocHTTPFetch
|
12
|
|
WSSGetListPageHtml
|
12
|
|
WSSSearch
|
11
|
|
WSSRSSList
|
10
|
|
WSSDispForm
|
9
|
|
WSSOutlookSync
|
7
|
The test results indicate that page-load times increase dramatically at bandwidths of 512 kilobytes per second (KBps) and lower, as demonstrated by the following graph.
.jpg)
-
The x-axis denotes response time, in seconds.
-
The y-axis denotes bandwidth, in megabits per second, and latency, in milliseconds.
-
Each line that is graphed represents a specific number of users and is based on a concurrency of 10 percent.
The following table lists the page-load test results for the common user operations:
| Bandwidth and latency | 500 | 1,000 | 5,000 | 10,000 | 100,000 |
|---|
|
45 (T3),
latency = 50 ms
|
< 1.5 sec
|
< 1.5 sec
|
< 1.5 sec
|
< 1.5 sec
|
< 1.5 sec
|
|
45 (T3),
latency = 100 ms
|
< 2 sec
|
< 2 sec
|
< 2 sec
|
< 2 sec
|
< 2 sec
|
|
45 (T3),
latency = 250 ms
|
< 4 sec
|
< 4.25 sec
|
< 4.25 sec
|
< 4.5 sec
|
< 4.5 sec
|
|
45 (T3),
latency = 500 ms
|
< 7.5 sec
|
< 7.75 sec
|
< 7.75 sec
|
< 8 sec
|
< 8 sec
|
|
12 (8 T1s),
latency = 50 ms
|
< 1.5 sec
|
< 1.5 sec
|
< 1.5 sec
|
< 1.5 sec
|
< 1.5 sec
|
|
12 (8 T1s),
latency = 100 ms
|
< 2.5 sec
|
< 2.5 sec
|
< 2.5 sec
|
< 2.5 sec
|
< 2.5 sec
|
|
12 (8 T1s),
latency = 250 ms
|
< 4.5 sec
|
< 4.5 sec
|
< 4.5 sec
|
< 4.5 sec
|
< 4.5 sec
|
|
12 (8 T1s),
latency = 500 ms
|
< 8 sec
|
< 8 sec
|
< 8 sec
|
< 8 sec
|
< 8 sec
|
|
6 (quad T1),
latency = 50 ms
|
< 2 sec
|
< 2 sec
|
< 2 sec
|
< 2 sec
|
< 2 sec
|
|
6 (quad T1),
latency = 100 ms
|
< 2.5 sec
|
< 2.5 sec
|
< 2.5 sec
|
< 2.5 sec
|
< 2.5 sec
|
|
6 (quad T1),
latency = 250 ms
|
< 5 sec
|
< 5 sec
|
< 5 sec
|
< 5 sec
|
< 5 sec
|
|
6 (quad T1),
latency = 500 ms
|
< 8 sec
|
< 8 sec
|
< 8 sec
|
< 8.5 sec
|
< 8.5 sec
|
|
3 (dual T1),
latency = 50 ms
|
< 3.5 sec
|
< 3.5 sec
|
< 4 sec
|
< 4 sec
|
< 4 sec
|
|
3 (dual T1),
latency = 100 ms
|
< 4 sec
|
< 4 sec
|
< 4 sec
|
< 4 sec
|
< 4 sec
|
|
3 (dual T1),
latency = 250 ms
|
< 4.5 sec
|
< 5 sec
|
< 5 sec
|
< 5 sec
|
< 5 sec
|
|
3 (dual T1),
latency = 500 ms
|
< 8 sec
|
< 8 sec
|
< 8 sec
|
< 8.5 sec
|
< 8.5 sec
|
|
1.5 (T1),
latency = 50 ms
|
< 6.5 sec
|
< 7.5 sec
|
< 7.5 sec
|
< 7.5 sec
|
< 7.5 sec
|
|
1.5 (T1),
latency = 100 ms
|
< 7 sec
|
< 7.5 sec
|
< 7.5 sec
|
< 7.5 sec
|
< 8 sec
|
|
1.5 (T1),
latency = 250 ms
|
< 7 sec
|
< 7.5 sec
|
< 7.5 sec
|
< 7.5 sec
|
< 7.5 sec
|
|
1.5 (T1),
latency = 500 ms
|
< 8.5 sec
|
< 8.5 sec
|
< 9 sec
|
< 9 sec
|
< 9 sec
|
|
0.5 (512 K),
latency = 50 ms
|
< 19 sec
|
< 19 sec
|
< 21 sec
|
< 21 sec
|
< 22 sec
|
|
0.5 (512 K),
latency = 100 ms
|
< 19 sec
|
< 19 sec
|
< 21 sec
|
< 21 sec
|
< 22 sec
|
|
0.5 (512 K),
latency = 250 ms
|
< 19 sec
|
< 19 sec
|
< 21 sec
|
< 21 sec
|
< 22 sec
|
|
0.5 (512 K),
latency = 500 ms
|
< 20 sec
|
< 20 sec
|
< 21 sec
|
< 22 sec
|
< 23 sec
|
Average page-load times for complex user operations
Complex user operations include downloading and uploading files and making other changes to content that require write operations.
Testing for this category was based on the following ratio of page requests:
| Page requested (read-write mix) | Percentage of overall requests |
|---|
|
WSSHomePage
|
25
|
|
WSSDocHTTPFetch
|
12
|
|
WSSGetListPageHtml
|
12
|
|
WSSDispForm
|
11
|
|
WSSSmallDocLibAllItems
|
9
|
|
WSSWebpartPage
|
8
|
|
WSSRSSList
|
7
|
|
WSSPutDocumentCoded
|
6
|
|
CreateTeamSite
|
4
|
|
WSSOutlookSync
|
4
|
|
WSSEditListItem
|
2
|
The test results indicate that operation times increase dramatically at bandwidths of 512 KBps and lower, as demonstrated by the following graph. Page-load times that fall to zero indicate that the operations failed to complete.
.jpg)
-
The x-axis denotes response time, in seconds.
-
The y-axis denotes bandwidth, in megabits per second, and latency, in milliseconds.
-
Each line that is graphed represents a specific number of users and is based on a concurrency of 10 percent.
The following table lists the page-load test results for the complex user operations:
.gif)
Note:
|
|---|
|
The acronym “DNF” indicates that the operation did not finish.
|
| Bandwidth and latency | 500 | 1,000 | 5,000 | 10,000 | 100,000 |
|---|
|
45 (T3),
latency = 50 ms
|
< 2 sec
|
< 2 sec
|
< 2.5 sec
|
< 2.5 sec
|
< 3.5 sec
|
|
45 (T3),
latency = 100 ms
|
< 3 sec
|
< 3 sec
|
< 3.5 sec
|
< 3.5 sec
|
< 4.5 sec
|
|
45 (T3),
latency = 250 ms
|
< 5 sec
|
< 5 sec
|
< 5.5 sec
|
< 6 sec
|
< 7 sec
|
|
45 (T3),
latency = 500 ms
|
< 8.5 sec
|
< 9.5 sec
|
< 10.5 sec
|
< 11 sec
|
< 13.5 sec
|
|
12 (8 T1s),
latency = 50 ms
|
< 2 sec
|
< 2.5 sec
|
< 3.5 sec
|
< 4.5 sec
|
< 5 sec
|
|
12 (8 T1s),
latency = 100 ms
|
< 3 sec
|
< 3.5 sec
|
< 3.5 sec
|
< 4 sec
|
< 5 sec
|
|
12 (8 T1s),
latency = 250 ms
|
< 5 sec
|
< 5 sec
|
< 5.5 sec
|
< 7 sec
|
< 7 sec
|
|
12 (8 T1s),
latency = 500 ms
|
< 8 sec
|
< 9 sec
|
< 9.5 sec
|
< 10.5 sec
|
< 11 sec
|
|
6 (quad T1),
latency = 50 ms
|
< 3 sec
|
< 4.5 sec
|
< 5 sec
|
< 5.5 sec
|
< 6.5 sec
|
|
6 (quad T1),
latency = 100 ms
|
< 3.5 sec
|
< 3.5 sec
|
< 5 sec
|
< 5.5 sec
|
< 6.5 sec
|
|
6 (quad T1),
latency = 250 ms
|
< 5 sec
|
< 5.5 sec
|
< 6 sec
|
< 6.5 sec
|
< 7 sec
|
|
6 (quad T1),
latency = 500 ms
|
< 8.5 sec
|
< 9 sec
|
< 10 sec
|
< 10.5 sec
|
< 11 sec
|
|
3 (dual T1),
latency = 50 ms
|
< 5.5 sec
|
< 7 sec
|
< 7.5 sec
|
< 8.5 sec
|
< 8.5 sec
|
|
3 (dual T1),
latency = 100 ms
|
< 5.5 sec
|
< 5.5 sec
|
< 8 sec
|
< 8 sec
|
< 11 sec
|
|
3 (dual T1),
latency = 250 ms
|
< 6 sec
|
< 6 sec
|
< 7 sec
|
< 9 sec
|
< 9.5 sec
|
|
3 (dual T1),
latency = 500 ms
|
< 9.5 sec
|
< 10 sec
|
< 10 sec
|
DNF
|
DNF
|
|
1.5 (T1),
latency = 50 ms
|
< 9 sec
|
< 10.5 sec
|
< 12 sec
|
< 13 sec
|
< 13.5 sec
|
|
1.5 (T1),
latency = 100 ms
|
< 10 sec
|
< 11 sec
|
< 12 sec
|
< 15 sec
|
< 15 sec
|
|
1.5 (T1),
latency = 250 ms
|
< 10 sec
|
< 12 sec
|
< 13 sec
|
< 13 sec
|
< 13.5 sec
|
|
1.5 (T1),
latency = 500 ms
|
< 10 sec
|
< 10 sec
|
< 12 sec
|
< 13 sec
|
< 15 sec
|
|
0.5 (512 Kbps),
latency = 50 ms
|
< 22 sec
|
< 22 sec
|
< 32 sec
|
< 33 sec
|
DNF
|
|
0.5 (512 Kbps),
latency = 100 ms
|
< 22 sec
|
< 23 sec
|
< 23 sec
|
< 42 sec
|
DNF
|
|
0.5 (512 Kbps),
latency = 250 ms
|
< 18.5 sec
|
< 26 sec
|
< 48 sec
|
DNF
|
DNF
|
|
0.5 (512 Kbps),
latency = 500 ms
|
< 24 sec
|
< 25 sec
|
DNF
|
DNF
|
DNF
|
Estimated time to crawl content
Crawl testing focused on performing crawls of the three different types of content sources with and without latency and bandwidth constraints. However, due to the time involved in setting up, running, and reinitializing each test scenario, the tests were only performed for one bandwidth and latency combination for each content source. With this in mind, testing focused on the most likely worst-case scenario at 100 ms of latency and 10 Mbps bandwidth.
SharePoint site content crawl
The following table reports the estimated time to crawl content in SharePoint sites based on available bandwidth, latency, and the volume of content.
| Bandwidth | 1 GB | 5 GB | 25 GB | 100 GB | 500 GB |
|---|
|
10 Mbps
Latency = none
Crawl rate = 467 MB/min
|
2 min
|
11 min
|
54 min
|
5 hr
30 min
|
17 hr
45 min
|
|
10 Mbps
Latency = 100 ms
Crawl rate = 330 MB/min
|
3 min
|
15 min
|
1 hr
15 min
|
5 hr
|
25 hr
15 min
|
File-share crawl
The following table reports the estimated time to crawl content in file shares based on available bandwidth, latency, and the volume of content.
| Bandwidth | 1 GB | 5 GB | 25 GB | 100 GB | 500 GB |
|---|
|
10 Mbps
Latency = none
Crawl rate = 467 MB/min
|
12 sec
|
1 min
|
5 min
|
20 min
|
1 hr
30 min
|
|
10 Mbps
Latency = 100 ms
Crawl rate = 330 MB/min
|
2 min
|
9 min
|
45 min
|
3 hr
|
15 hr
|
HTTP crawl
The following table reports the estimated time to crawl content across HTTP sites based on available bandwidth, latency, and the volume of content.
| Bandwidth | 1 GB | 5 GB | 25 GB | 100 GB | 500 GB |
|---|
|
10 Mbps
Latency = none
Crawl rate = 467 MB/min
|
79 min
|
6 hr
30 min
|
33 hr
|
132 hr
|
659 hr
|
|
10 Mbps
Latency = 100 ms
Crawl rate = 330 MB/min
|
3 hr
|
15 hr
15 min
|
76 hr
30 min
|
305 hr
40 min
|
1529 hr
|
Lab environment and WAN testing tools
This section shows the specific characteristics of the lab environment that was used to produce the test results. It also lists specific tools that were used.
Network storage area configuration
EMC CLARiiON CX700 for storage was configured as follows:
-
(2) Service process controllers with a 4-GB cache each
-
(30) 73-GB spindles at 10,000 RPM with fiber channel backbone
-
Brocade SilkWorm 3850 fiber channel switch
SQL Server configuration
-
Hardware: Dell PowerEdge 6950
-
OS: Windows Server 2003 R2 Enterprise x64 Edition with SP2
-
CPU: (4) Dual-Core AMD Opteron 8218 processors (2.60 GHz)
-
Memory: 32 GB of RAM
-
Controller: (2) QLogic 4-GB fiber channel adapters
-
Database: SQL Server 2005 with Service Pack 2 (SP2) (x64)
-
Drive X: 100 GB (TempDB)
-
Drive Y: 100 GB (logs)
-
Drive Z: 300 GB (data)
Index server configuration
-
Hardware: Dell PowerEdge 6850
-
OS: Windows Server 2003 R2 Enterprise x32 Edition with SP2
-
CPU: (4) x 3.0 GHz Dual-Core Intel Xeon processors
-
Memory: 32 GB of RAM
Front-end Web server configuration
-
Hardware: Dell PowerEdge 2950
-
OS: Windows Server 2003 R2 Enterprise x32 Edition with SP2
-
CPU: (2) x 3.0 GHz Dual-Core Intel Xeon processors
-
Memory:16 GB of RAM
WAN accelerator configuration
The WAN accelerator device was the Shunra VE STN.
Load generation tools
Download this book
See Also