Intel 170 Servers User Manual
The chart above shows the CPW efficiency of operations (larger is better). Note the CPW per Mbits/sec
scale on the left - as it’s different for each chart.
scale on the left - as it’s different for each chart.
For an IXS or IXA, the port-based VE has the least CPW or smaller packets due to consolidation of
transfers available in Licensed Internal Code. The VLAN-based transfers have the greatest cost (However
the total would be split during inter-LPAR communications).
transfers available in Licensed Internal Code. The VLAN-based transfers have the greatest cost (However
the total would be split during inter-LPAR communications).
For iSCSI, the cost of using standard frames is 1.5 to 4.5 times higher than jumbo frames.
17.6.3 Windows CPU Cost
The next chart illustrates the cost of iSCSI port-based, and VLAN-based virtual ethernet operations on a
windows CPU. In this case, the CPUs used are 3.2Ghz Xeon uniprocessor between HS20 BladeCenter
servers. This cost is compared to operations across the external gigabit NIC connections. Again, the
jumbo frames operations are less expensive than standard frames, though the external NIC is twice as
efficient in General.
windows CPU. In this case, the CPUs used are 3.2Ghz Xeon uniprocessor between HS20 BladeCenter
servers. This cost is compared to operations across the external gigabit NIC connections. Again, the
jumbo frames operations are less expensive than standard frames, though the external NIC is twice as
efficient in General.
Streaming TCP Total Windows CPU Util
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
128
25
6
512
10
24
20
48
40
96
81
92
16
384
32
76
8
Transaction Size (bytes)
To
ta
l C
P
U
U
til
pe
r M
B
it
s
/s
e
c
External - Window s to Window s (std)
External - Window s to Window s (jumbo)
Port Based VE - Window s to Window s (std)
Port Based VE - Window s to Window s (jumbo)
Vlan Based VE - Window s to Window s (std)
Vlan Based VE - Window s to Window s (jumbo)
External - Window s to Window s (jumbo)
Port Based VE - Window s to Window s (std)
Port Based VE - Window s to Window s (jumbo)
Vlan Based VE - Window s to Window s (std)
Vlan Based VE - Window s to Window s (jumbo)
17.7 File Level Backup Performance
The Integrated Server support allows you to save integrated server data (files, directories, shares, and the
Windows registry) to tape, optical or disk (*SAVF) in conjunction with your other i5/OS data. That is,
this “file level backup” approach saves or restores the Windows files on an individual basis within the
stream of other i5/OS data objects. It’s not recommended that this approach is used as a primary backup
procedure. Rather, you should still periodically save your NWS storage spaces and the NWSD associated
with your Windows server or Linux server for disaster recovery.
Windows registry) to tape, optical or disk (*SAVF) in conjunction with your other i5/OS data. That is,
this “file level backup” approach saves or restores the Windows files on an individual basis within the
stream of other i5/OS data objects. It’s not recommended that this approach is used as a primary backup
procedure. Rather, you should still periodically save your NWS storage spaces and the NWSD associated
with your Windows server or Linux server for disaster recovery.
Saving individual files does not operate as fast as saving the individual storage spaces. The save of a
storage space on a equivalent machine and tape is about 210 Gbytes per hour, compared to the
approximately 70 Gbytes per hour achieved using iSCSI below.
storage space on a equivalent machine and tape is about 210 Gbytes per hour, compared to the
approximately 70 Gbytes per hour achieved using iSCSI below.
The chart below compares some SAV and RST rates for iSCSI and an IXA attached server. These results
were measured on a System i Model 570 2-way 26F2 processor (7495 capacity card). The i5/OS release
was V5R4. The IXA attached comparison server was an x365 xSeries (4way 2.5Ghz Xeon with IXA and
Windows Server 2003 with SP1). The iSCSI servers were HS20 BladeCenter servers with a copper
iSCSI (p/n 26K6489) daughter card. Switches were Nortel L2/3 Ethernet (p/n 26K6524)
were measured on a System i Model 570 2-way 26F2 processor (7495 capacity card). The i5/OS release
was V5R4. The IXA attached comparison server was an x365 xSeries (4way 2.5Ghz Xeon with IXA and
Windows Server 2003 with SP1). The iSCSI servers were HS20 BladeCenter servers with a copper
iSCSI (p/n 26K6489) daughter card. Switches were Nortel L2/3 Ethernet (p/n 26K6524)
.
The target tape
drive was a model 5755-001 (Ultrium LTO 2). All tests were run with jumbo frames enabled.
IBM i 6.1 Performance Capabilities Reference - January/April/October 2008
©
Copyright IBM Corp. 2008
Chapter 17 - Integrated BladeCenter and System x Performance
291