HP ProLiant 4500 Disk Subsystem Performance and Scalability - Page 15

Table 7, Host Bus Memory Transfer Rates

Get HP ProLiant 4500 PDF manuals and user guides View all HP ProLiant 4500 manuals
Add to My Manuals
Save this manual to your list of manuals

Page 15 highlights

WHITE PAPER (cont.) Note: Compaq uses coalescing algorithms to optimize disk performance. ECG025.0997 ... Table 7 lists the Host bus transfer rates for the following Compaq servers: Table 7: Host Bus (Memory) Transfer Rates Server Name Transfer Rate ProLiant 5000, 6000, 6500 and 7000 540 MB/s ProLiant 1500, 2500 and 4500 267 MB/s The last example discussed how READ performance is increased. Now let's discuss how WRITE performance is enhanced on a system by taking advantage of posted writes. Posted writes take place when file system or disk controller caching temporarily holds one or more blocks of data in memory until the hard disk is not busy. The system then combines or "coalesces" the blocks of data into larger blocks and writes them to the hard disk. This results in fewer and larger sequential I/Os. For example, a network server is used to store data. This server is responsible for completing hundreds of client requests. If the server happened to be busy when data was being saved, the server's file system cache tells the application that the data has been saved so that the application can continue immediately without having to wait for the disk I/O to complete. Coalescing is also commonly referred to as "Elevator seeking." This coined phrase became popular because it provides the perfect analogy for describing coalescing. For instance, an elevator picks up and drops off passengers at their requested stop in the most efficient manner possible. If you were on level 6, the elevator on level 2, and other passengers on levels 1 and 7, the elevator would first stop on level 1 to pick up the passenger going up. Next, the elevator would stop on level 6, then 7 and then take everyone to level 9, their destination. The elevator would not perform all of the requests individually, instead it reorders then completes those requests in a more efficient manner. This same analogy applies to coalescing when writing data to different sectors on a disk. As an example, Joe saves or "writes" data B to the hard disk, then he saves data A to the same disk. And finally, he saves data C as well. Instead of completing 3 separate I/Os for B, A, then C, the system reorders the write requests to reflect data ABC then performs a single sequential I/O to the hard disk, thus improving disk performance. 15

  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 10
  • 11
  • 12
  • 13
  • 14
  • 15
  • 16
  • 17
  • 18
  • 19
  • 20
  • 21
  • 22
  • 23
  • 24
  • 25
  • 26
  • 27
  • 28
  • 29
  • 30
W
HITE
P
APER
(cont.)
15
ECG025.0997
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Table 7 lists the Host bus transfer rates for the following Compaq servers:
Table 7:
Host Bus (Memory) Transfer Rates
Server Name
Transfer Rate
ProLiant 5000, 6000, 6500 and 7000
540 MB/s
ProLiant 1500, 2500 and 4500
267 MB/s
The last example discussed how READ performance is increased.
Now let’s discuss how
WRITE performance is enhanced on a system by taking advantage of posted writes.
Posted writes take place when file system or disk controller caching temporarily holds one
or more blocks of data in memory until the hard disk is not busy.
The system then
combines or “coalesces” the blocks of data into larger blocks and writes them to the hard
disk.
This results in fewer and larger sequential I/Os.
For example, a network server is
used to store data.
This server is responsible for completing hundreds of client requests.
If the server happened to be busy when data was being saved, the server’s file system
cache tells the application that the data has been saved so that the application can
continue immediately without having to wait for the disk I/O to complete.
Coalescing is also commonly referred to as “Elevator seeking.”
This coined phrase
became popular because it provides the perfect analogy for describing coalescing.
For
instance, an elevator picks up and drops off passengers at their requested stop in the
most efficient manner possible.
If you were on level 6, the elevator on level 2, and other
passengers on levels 1 and 7, the elevator would first stop on level 1 to pick up the
passenger going up.
Next, the elevator would stop on level 6, then 7 and then take
everyone to level 9, their destination.
The elevator would not perform all of the requests
individually, instead it reorders then completes those requests in a more efficient manner.
This same analogy applies to coalescing when writing data to different sectors on a disk.
As an example, Joe saves or “writes” data B to the hard disk, then he saves data A to the
same disk.
And finally, he saves data C as well.
Instead of completing 3 separate I/Os
for B, A, then C, the system reorders the write requests to reflect data ABC then performs
a single sequential I/O to the hard disk, thus improving disk performance.
Note:
Compaq uses coalescing
algorithms to optimize disk
performance.