The elements used for the IMS testing were P-CSCF, S- CSCF, I-CSCF and HSS. IMS Core is based on the Fokus [7] IMS Core OpenSource program development project. HSS is based on the MySQL [6] database. The IMS uses database via Java based Focus Home Subscriber Server (FHoSS). In the SIP network side, the OpenSIPS [8] (version 1.5.3) server was used. It used the same MySQL database as the IMS but via its own module. The basic services in the test network were VoIP, Video calls and Instant Messaging (IM). Additionally there was Video on Demand. All services were located in separate Application Servers (AP). The research network was connected to the Internet through a Finnish Funet connection enabling IP based VoIP/Video calls all over the Internet. The hardware configuration had the following characteristics: X-CSCF servers and the HSS were located in the same virtualization based Linux PC framework: there was a server Dell PowerEdge 2950 with 2 Intel Xeon E5420 (Quad Core, 2,50 GHz) -processors and 32 GB ECC-DDR2 memory,and 4 750 GB SAS disks using RAID 10 configuration. The server
platform was also equipped with 6 1 Gb Ethernet connections. Virtualization was done with the Citrix’s XenServer 5.0.0 (Hotfix 3). The virtual machine was configured with 2 virtual processors, each using a 4096 MB of the main memory, 15 GB of disk memory and one 1Gb Ethernet port. However, due to virtualization problems the Ethernet connection acted only as a 100 Mb connection. The priority level for each of
the virtual processors was 65536/65536 (the highest possible). The operating system used was Debian GNU/Linux 5.0 (lenny, amd64). As the measurement program we used the IMS Bench SIPp program version 340(27.11.2007) runtime in the virtual
machines. The actual measurement platform was based on Intel Core 2 Duo E8400 CPU and with 4 GB DDR2 memory and a 1Gb Ethernet port. The operating system used was Debian GNU/Linux 5.0 (squeeze, amd64) with kernel V2.6.30. The kernel Timer Frequency value was changed from original 250 Hz to 1000 Hz that improves the measurement accurancy to 1 msec.
The elements used for the IMS testing were P-CSCF, S- CSCF, I-CSCF and HSS. IMS Core is based on the Fokus [7] IMS Core OpenSource program development project. HSS is based on the MySQL [6] database. The IMS uses database via Java based Focus Home Subscriber Server (FHoSS). In the SIP network side, the OpenSIPS [8] (version 1.5.3) server was used. It used the same MySQL database as the IMS but via its own module. The basic services in the test network were VoIP, Video calls and Instant Messaging (IM). Additionally there was Video on Demand. All services were located in separate Application Servers (AP). The research network was connected to the Internet through a Finnish Funet connection enabling IP based VoIP/Video calls all over the Internet. The hardware configuration had the following characteristics: X-CSCF servers and the HSS were located in the same virtualization based Linux PC framework: there was a server Dell PowerEdge 2950 with 2 Intel Xeon E5420 (Quad Core, 2,50 GHz) -processors and 32 GB ECC-DDR2 memory,and 4 750 GB SAS disks using RAID 10 configuration. The serverplatform was also equipped with 6 1 Gb Ethernet connections. Virtualization was done with the Citrix’s XenServer 5.0.0 (Hotfix 3). The virtual machine was configured with 2 virtual processors, each using a 4096 MB of the main memory, 15 GB of disk memory and one 1Gb Ethernet port. However, due to virtualization problems the Ethernet connection acted only as a 100 Mb connection. The priority level for each ofthe virtual processors was 65536/65536 (the highest possible). The operating system used was Debian GNU/Linux 5.0 (lenny, amd64). As the measurement program we used the IMS Bench SIPp program version 340(27.11.2007) runtime in the virtualmachines. The actual measurement platform was based on Intel Core 2 Duo E8400 CPU and with 4 GB DDR2 memory and a 1Gb Ethernet port. The operating system used was Debian GNU/Linux 5.0 (squeeze, amd64) with kernel V2.6.30. The kernel Timer Frequency value was changed from original 250 Hz to 1000 Hz that improves the measurement accurancy to 1 msec.
การแปล กรุณารอสักครู่..
