Solutions Business Unit Compaq Computer Compaq ProLiant Servers Corporation CONTENTS This white paper provides guidelines for planning the configuration of Compaq ProLiant Introduction to the 2500 and ProLiant 7000 computers that run Microsoft Site Server Internet Locator Microsoft Site Server and Personalization System. In order to determine capacity planning guidelines, Server ......
Page 2
This test is not a determination of product quality or correctness, nor does it ensure compliance with any federal, state or local requirements. Compaq does not warrant products other than its own strictly as stated in Compaq product warranties.
HITE APER (cont.) N T R O D U C T I O N T O I C R O S O F T I T E E R V E R As enterprise customers become more dependent on the Intranet, Internet, and World Wide Web, there is a tendency to get bogged down in the creation and management of web-based applications.
Page 4
HITE APER (cont.) through command-line tools or through the Web browser. This allows for the automatic updating of Web farms and makes the CRS a vital tool for the on-line publisher. Internet Locator Server allows you to find others who are currently on line. It finds information that is constantly changing, such as a user’s IP address, and continually updates its information as users log on and off.
X E C U T I V E U M M A R Y This white paper provides guidelines for planning the configuration of Compaq ProLiant 2500 and ProLiant 7000 computers to run two features from Microsoft Site Server: Internet Locator Server (ILS) and Personalization System (PS).
HITE APER (cont.) E R F O R M A N C E E A S U R E M E N T S The goal of capacity planning is to determine the types and configurations of computer systems, networks, and other resources that can support a specified number of users with a precise response time.
PS testing Memory 256 MB EDO ECC Network 1 x Dual Port Compaq NetFlex3, 100Base-TX Disk 4 x 4.3 GB 10,000 RPM disks in a RAID 5 configuration with a SMART-2 controller Internet Locator Server (ILS) Performance Measurement Procedures For the ILS performance measurements, both InetLoad and perfmon were used.
Page 8
HITE APER (cont.) Table 2: ILS Operations Measured ILS Operation Definition and Comments Connect Connection to the ILS directory service. This operation must be done before any application can use ILS and is always done at least once by all ILS applications. The performance of the LDAP command sequence CONNECT, BINDSIMPLE ANONYMOUS, and QUIT was measured to simulate this operation.
Page 9
HITE APER (cont.) ILS Test Procedures The basic procedure used for performing the ILS tests was: 1. Automatically start IIS and the LDAP server at SUT boot time. This enables both IIS and LDAP logging. 2. Use the ASP-enabled Web page to create a simulated online user base of the appropriate size. 3.
Page 10
HITE APER (cont.) 1800 1539 1532 1512 1485 1600 1400 1200 1360 1328 1276 1000 1000 5000 10000 Simulated Online Users Connect Direct Lookup Figure 2: ProLiant 7000 ILS Connect, Add, and Direct Lookup Operations Performance Summary In Tables 4 through 9, the “Available Memory” perfmon measurement is the last sample taken during the test instead of the average.
Page 12
HITE APER (cont.) the type of CPU in the system, its speed, the number of processors, and how much of the CPU is utilized the type of network, its speed, the number of Network Interface Cards (NICs), and the device drivers for the NICs the amount and type of memory in the system the operating system...
HITE APER (cont.) Operating Not enough Unless ILS implementation details are known, you cannot System information determine whether performance bottlenecks are in the operating system or ILS. ILS Software Not enough Unless ILS implementation details are known, you cannot Information determine whether performance bottlenecks are in the operating system or ILS.
Page 14
HITE APER (cont.) Table 12: PS User Scenario Operation Operation # Operation Frequency Comment Create a new UPD entry 5% of the users will be new, so create a new UPD entry for each of them. Get a personalized Web page Each Web page will consist of an HTTP GET of the .asp file and 4 HTTP GETs of .gif files.
HITE APER (cont.) Table 13: ProLiant 2500 PS Performance Measurements Measurement 1,000 Users 10,000 Users 100,000 Users 1,000,000 Users CPU Utilization 96.7% 91.5% 85.0% Web pages requested/sec 49.6 43.5 37.2 32.1 Average Connect Time (sec) ASP requests/sec 53.5 47.3 40.3 35.0 UPD reads/sec 53.5...
Page 17
HITE APER (cont.) Network The total network traffic is less than 15% of the available bandwidth. Memory There is between 87 MB and 96 MB of memory available. The “Pages/sec” measurement shows a high page fault rate with larger UPDs. The cause is most likely that PS is using memory-mapped files.
HITE APER (cont.) ProLiant 7000 PS Bottlenecks Analysis of PS bottlenecks on the ProLiant 7000 shows: Table 16: ProLiant 7000 Analysis of PS Bottlenecks Resource Bottleneck Comments The CPU utilization decreases as the number of users in the UPD increases. Network The total network traffic is less than 15% of the available bandwidth.
Page 19
HITE APER (cont.) Figure 6 shows the test lab configuration used for the ProLiant 7000 testing. The number of client systems and networks was also determined by experimentation until the CPUs were 100% utilized. Table 18 shows the actual number of client systems used and the configurations for each test.
By doing a “what-if” analysis using this workbook, which Compaq computers, and how many are needed to support the specified number of users with a given response time, can be determined.
Page 21
HITE APER (cont.) Be sure that the network bandwidth is sufficient to handle the projected transaction rate. If the server will use a 100 Mbit/second network, bandwidth will not be a bottleneck. However, at least two 10 Mbit/second networks will be needed to handle the maximum transaction rate supported by either a ProLiant 2500 or ProLiant 7000.
System. ILS and PS performance measurements for both the ProLiant 2500 and 7000 have shown that they were capable of generating loads that fully utilize their CPUs. Microsoft Site Server, on Compaq ProLiant 2500 and 7000, delivers the return on investment businesses are looking for from their Intranet and Internet sites.
An Excel workbook is included in the online version of this White Paper for your use in determining how many of your users can use Internet Locator Server in Compaq ProLiant 2500 and ProLiant 7000 servers configured like those tested in this White Paper.
Page 24
HITE APER (cont.) (This must include the number of refresh operations that keep the user active. A refresh will typically happen every 10 minutes; however, this may vary by application and configuration.) 2. Below are the number of users that can be supported by a ProLiant 2500 and a ProLiant 7000 ProLiant 2500 ILS User Capacity A.
Page 25
HITE APER (cont.) C. Number of PS-enabled Web pages your 43.5 server can deliver per second (This number is usually determined by benchmarking. To get an estimate for a ProLiant 2500 enter 43.5 and for a ProLiant 7000 enter 57.3. The accompanying white paper will give you other values for these systems.
HITE APER (cont.) B: G P P E N D I X L O S S A R Y The following terms are used in this white paper: ASP: Active Server Page, a feature of Microsoft Internet Information Server Clients: The number of processes or threads simultaneously requesting Web services from the server.